sensors

Article
Short-Term Traffic State Prediction Based on the
Spatiotemporal Features of Critical Road Sections
Gang Yang 1 , Yunpeng Wang 1,2 , Haiyang Yu 1,2, *, Yilong Ren 1,2,3 and Jindong Xie 4
1 School of Transportation Science and Engineering, Beijing Key Laboratory for Cooperative Vehicle
Infrastructure System and Safety Control, Beihang University, Beijing 100191, China;
yanggang@buaa.edu.cn (G.Y.); ypwang@buaa.edu.cn (Y.W); yilongren@buaa.edu.cn (Y.R.)
2 Beijing Advanced Innovation Center for Big Data and Brain Computing, Beihang University,
Beijing 100191, China
3 Jiangsu Province Collaborative Innovation Center of Modern Urban Traffic Technologies, Sipailou 2,
Nanjing 210096, China
4 School of Electronic and Information Engineering, Beihang University, Beijing 100191, China;
jindong.xie@buaa.edu.cn
* Correspondence: hyyu@buaa.edu.cn; Tel.: +86-186-1245-2770




Received: 17 May 2018; Accepted: 10 July 2018; Published: 14 July 2018


Abstract: Recently, short-term traffic prediction under conditions with corrupted or missing data
has become a popular topic. Since a road section has predictive power regarding the adjacent roads
at a specific location, this paper proposes a novel hybrid convolutional long short-term memory
neural network model based on critical road sections (CRS-ConvLSTM NN) to predict the traffic
evolution of global networks. The critical road sections that have the most powerful impact on
the subnetwork are identified by a spatiotemporal correlation algorithm. Subsequently, the traffic
speed of the critical road sections is used as the input to the ConvLSTM to predict the future traffic
states of the entire network. The experimental results from a Beijing traffic network indicate that
the CRS-ConvLSTM outperforms prevailing deep learning (DL) approaches for cases that consider
critical road sections and the results validate the capability and generalizability of the model when
predicting with different numbers of critical road sections.

Keywords: short-term traffic prediction; structural missing data; deep learning; critical road sections;
spatiotemporal correlation

1. Introduction
Real-time traffic state prediction plays a vital role in traffic management and public service.
The ability to timely, accurately and efficiently predict the evolution of traffic can assist travelers and
government agencies in reacting to possible congestion ahead of time. In recent decades, reliance on
efficient access to massive traffic data, such as loop detectors and floating cars carrying global
positioning system (GPS) devices, has provided an impetus for a considerable amount of data-driven
computational approaches to estimate and forecast future traffic states. Compared with data-driven
methods, mathematical or statistical models that are derived from macroscopic and microscopic
theories of traffic flow exhibit difficulties addressing unstable traffic conditions and complex road
settings due to their strong hypotheses and assumptions [1].
Data-driven methods, including the autoregressive integrated moving average model (ARIMA),
support vector machine (SVM), Bayesian network, neural network (NN) and so forth, have achieved
promising results due to their greater potential in processing complex nonlinear problems [2].
Among all the data-driven methods, the deep learning approach has been validated for its efficiency
since it is capable of exploiting much deeper architectures and processing high-dimensional sets of

Sensors 2018, 18, 2287; doi:10.3390/s18072287 www.mdpi.com/journal/sensors

Sensors 2018, 18, 2287 2 of 26

explanatory variables in transportation networks [3]. However, the veracity of the data (i.e., noisy and
missing data) in the real world affects predictive accuracy to a great extent. Especially for taxi-based
floating car data (FCD), spatial coverage relies heavily on the penetration rate of probing taxis [4].
The use of lower density probes on certain roads will result in a large variance of traffic speed that
does not accurately reflect real traffic conditions. This problem often occurs on minor roads and
leads to structurally unreliable traffic data. In addition, urban tall buildings and other large objects
usually block or strongly degrade GPS signals to invalidate road traffic data [5]. These setbacks caused
a long-term loss of traffic data on some road sections, namely structural missing data, which weakens
the predictive accuracy and limits the ability to provide a practical and reliable forecasting result.
This paper seeks to provide a solution to settle these matters.
The influence between downstream and upstream areas has been studied for a long period
of time [6]. Recently, researchers have found that previous conditions of a road section have
an impact on the current state of its adjacent roads at a specific local network [7]. This conclusion
indicates that the future traffic conditions of a road can be predicted by the current state of its
neighbors. The predictive power between two adjacent roads can be estimated by their spatiotemporal
correlation [8]. Therefore, if we extract data on the critical road sections that have the most dominant
predictive power, we can characterize the spatiotemporal features of traffic flow and predict the future
traffic conditions of the overall network. In this way, the traffic conditions of minor roads affected by
structural missing or unreliable data can be predicted by their critical adjacent roads.
In this paper, we propose a novel hybrid convolutional long short-term memory neural network
model based on critical road sections (CRS-ConvLSTM NN) to predict future traffic conditions in the
overall network. With the assistance of a spatiotemporal correlation algorithm (STCA), the critical road
sections that have a significant impact on the local traffic networks are identified and extracted from
the network. First, the spatial relevance among road sections is evaluated by establishing a spatial
adjacency matrix. The matrix is constructed according to the topological relationships of the urban
network. Then, the strength of the influence between a given road section and its neighbors under
conditions of a 2 min to 1 h time delay can be calculated based on the correlation theory. The degree of
influence can be adopted as a criterion for evaluating the importance of the roads. Because the value of
the spatiotemporal correlation is different for each delay of each road section, the technique for order
preference by similarity to an ideal solution (TOPSIS) is applied to synthetically consider its impact on
the subnetwork by ranking their orders. Then, a certain proportion of road sections is regarded as the
critical sections according to their ranks.
In the prediction section, we introduce the ConvLSTM NN and show that it outperforms
other models, such as artificial neural networks (ANNs), convolutional neural networks (CNNs),
long short-term memory neural networks (LSTMs) and the stacked autoencoder model (SAE).
The number of critical road sections we extracted from all the road sections affects the forecast
performance in terms of the loss function of the CRS-ConvLSTM model. To the best of our
knowledge, this paper presents the first report of a method for predicting the urban traffic evolution
on a network-wide scale with partial sections that characterize the spatiotemporal features of the
integral network.
The contribution of the CRS-ConvLSTM model lies in addressing structural data failure problems
by exploiting the potential of the predictive power among sections. With the preprocessing strategies
exerted on the input data based on the STCA, our model can clearly reflect the spatial and temporal
characteristics of the road network. Practical experiments on real-life speed data in a large-scale
network of Beijing City demonstrate that the proposed CRS-ConvLSTM model can precisely predict
the future traffic network states. Additionally, the fluctuation of performance under different numbers
of critical road sections and various situations (e.g., stochastic and extreme case) is also tested.
The remainder of our paper is outlined as follows. Section 2 discusses the literature on real-time
short-term traffic prediction. Section 3 elaborates on the spatiotemporal correlation algorithm to
determine the degree of impact of the road sections and proposes a method of identifying the critical
Sensors 2018, 18, 2287 3 of 26

road sections. Section 4 constructs the CRS-ConvLSTM model to predict the speed of the overall
traffic network based on the data of critical road sections. To validate the effectiveness of the proposed
approach, experiments on a large-scale traffic network of Beijing City are conducted. Section 5 evaluate
the performances under different numbers of critical road sections based on the predictive accuracy.
The final section presents the conclusions and discusses future work.

2. Literature Review
In recent decades, a considerable number of short-term traffic forecast models have been
developed to predict future traffic states ranging from a few seconds to a few hours ahead.
The approaches can be generally categorized into parametric and nonparametric approaches [3,9].
The parametric approaches mainly focus on predetermining the structure of the model based on
theoretical or physical assumptions and then tuning a set of parameters to faithfully reflect the evolution
of traffic conditions in the real world [3,9]. Among all the parametric approaches, model-based
time-series analyses, especially ARIMA, are widely used for traffic predictions. Since Ahmed and Cook
first introduced the ARIMA model in univariate single-point settings [10], a number of variations,
such as seasonal ARIMA [11] and STARIMA [12], were proposed to improve prediction accuracy.
However, the majority of parametric approaches fail to provide a favorable result under unstable traffic
conditions, such as unexpected events or extreme weather [3]. To address this problem, numerous
nonparametric approaches have been proposed and they present a number of advantages, such as
the ability to be free from strong assumptions underlying the model and to adaptively learn the
implicit dynamic traffic characteristics through archived traffic data; these approaches include Kalman
filters [13–15], SVM [16–18], NNs [19–22] and hybrid methods [23,24].
Vlahogianni et al. indicated that artificial intelligence (AI) has gradually become the dominant
approach for transportation prediction because of its inherent propensity for hypothesis-free techniques
that are capable of addressing multidimensional and nonlinear problems [3]. With the strength of
flexibility, adaptability, learning and generalizability, NNs have received tremendous interest and been
extensively used as a capable alternative to address transportation problems. Smith and Demetsky
established back-propagation neural networks (BPNNs) to forecast traffic volume and the results clearly
showed that the BPNNs outperformed the historical average and ARIMA models [21]. Chen and
Grant-Muller investigated the dynamic characteristics of a neural network and constructed dynamic
NNs in cooperation with varied and complex traffic patterns [25]. Huang et al. constructed a NNs
model to predict traffic speed by considering weather conditions, specifically the visibility, temperature
and moisture [26]. Another line of research on short-term traffic prediction incorporated traditional
NN approaches with other computational intelligence (CI) approaches or statistical methods since
the prediction that depended exclusively on NNs lacked an optimal generalization capability [27].
Chan et al. adopted a hybrid exponential smoothing method and Levenberg-Marquardt algorithm
(EXP-LM) to remove the lumpiness that weakens the generalization capability of the approach from
traffic data and then trained the model with NNs [27]. With the purpose of avoiding arbitrariness
when selecting a suitable structure of a network, Hinsbergen et al. calculated an evidence factor based
on a Bayesian committee to facilitate the integration of different NNs and it obviously outperformed
a single NN when applied to an empirical experiment [28].
Recently, deep learning approaches capable of processing nonlinear high-dimensional problems
have been increasingly developed; these approaches are capable of addressing complex traffic
flow characterized by both spatial and temporal features. Different from other traditional ANNs
models, deep learning algorithms extract inherent features of the data using multiple layers or deep
architectures and thus achieve better results [29]. Under the guidance of applications developed
in the domains of image, video, audio and language learning tasks [30], deep learning approaches
have proven their effectiveness for traffic predictions. In 2014, Huang et al. employed a deep belief
network (DBN) with multitask learning for traffic flow predictions [2]. Lv et al. proposed an SAE
and its performance is superior to that of BPNNs, the random walk (RW) forecast method, SVMs and
Sensors 2018, 18, 2287 4 of 26

the radial basis function (RBF) NN at different prediction horizons [31]. Furthermore, the recurrent
neural network (RNN), which presents the intrinsic properties of creating and processing memories
of arbitrary sequences of input patterns, is regarded as an excellent model for capturing the time
sequential features of traffic flow, especially the LSTM network because it overcomes the vanishing and
exploding gradient problems when capturing long temporal dependency [32–35]. Another promising
branch of NNs that has been widely adopted for traffic forecasting is the CNN. Ma et al. viewed the
traffic conditions as sequential images and then predicted traffic speeds after 10 min and 20 min based
on the spatiotemporal features of the network [36].
Furthermore, numerous efforts have been devoted to emphasizing the temporal characteristics
and spatial dependencies of predictions [3]. Min et al. proposed a multivariate spatial-temporal
autoregressive (MSTAR) model that defined spatial correlations among links by the reachability
of one link was reachable by another in a certain time interval [8]. Cai et al. predicted multistep
traffic conditions using a k-nearest neighbor model based on a spatiotemporal state matrix [37].
Habtemichael et al. predicted future traffic through an enhanced K-NN method. A similar historical
dataset was identified as the forecasting reference by means of the weighted Euclidean distance and the
rank-exponent method [38]. In recent years, deep learning approaches, especially CNNs and LSTMs,
have contributed to capturing spatiotemporal features since the evolution of traffic conditions shares
certain similarities in the spatial and temporal dimensions with video, speech and so forth. Wu et al.
extracted spatial and temporal features by a CNN and LSTM, respectively and predicted traffic volume
combining these two approaches [39]. The results showed that the convolutional LSTM structure
can provide a potential alternative for traffic predictions. Wang et al. proposed an eRCNN model
that integrated spatiotemporal features and then trained the recurrent CNN by reducing feedback
predictive errors [40]. Yu et al. depicted the traffic network as a grid-based diagram, with each grid
representing the traffic speed in a certain location [41]. Then, SRCNs that integrated the CNN with
LSTM were utilized to forecast the speed of a large-scale network and the results outperformed other
deep learning approaches in terms of the mean absolute percentage error (MAPE) and root mean
squared error (RMSE).
In summary, since a wide spectrum of short-term prediction models have been applied in traffic
forecasting, selecting the appropriate modeling approach for a specific circumstance is crucial for
generating accurate and reliable prediction results. Among all the parametric and nonparametric
approaches proposed for short-term traffic forecasting, deep learning methods have emerged as the
alternative with the most potential due to their remarkable abilities for addressing nonlinear and
high-dimensional problems. However, few prediction methods can achieve satisfactory accuracy under
the conditions of structural missing data. Conventional techniques that compensate for incomplete
data rely on filling in missing values with reasonable replacements before feeding the data into the
prediction model [42]. Our method represents a new concept for addressing structural missing data
problems. Moreover, because the inherent spatiotemporal characteristics of traffic flow affect the
predictive performance to a great extent, various studies have concentrated on incorporating temporal
and spatial dependencies into the modeling phase. In this paper, we propose the STCA to extract
critical road sections that have the most impact and predictive power for future traffic states and
future traffic conditions of the overall large-scale network are predicted based on the CRS-ConvLSTM.
Essentially, the model employs partial roads to forecast the traffic states of the entire network, which is
distinct from most existing prediction approaches. Furthermore, the impact of structural missing data
on the performance of the model is explored.

3. Methodology
The study focuses on utilizing data from critical road sections to predict future traffic conditions
of an overall urban transportation network. Specifically, the critical road sections are identified by
a spatiotemporal correlation algorithm that simultaneously presents the spatial and temporal features
Sensors 2018, 18, 2287 5 of 26

of the traffic network. In this paper, a ConvLSTM NN that combines a CNN and LSTM is used to
predict the future short-term traffic speed of the network.

3.1. Identifying Critical Road Sections by Spatiotemporal Features

3.1.1. Constructing Spatial Weights Matrix

A spatial weights matrix represents the spatial dependency among road sections in a traffic
network. According to graph theory [43], a traffic network can be abstracted to a directed graph
G = (N, L) composed of N nodes and L links. Certain nodes of the traffic network are not directly
connected through a path road. According to Tobler’s First Law [44], near traffic flows are more closely
related than distant flows [45]. To exploit the static and physical properties of the transportation
network, we construct a k-order spatial weights matrix through the topological adjacency relationships
between road sections. If edges i and j are connected directly or indirectly, we call these two edges the
“kth order neighbors” and their adjacency relationship can be expressed as Equation (1):
(
(k) 1, when adjacent
eij = (1)
0, others

(k)
The weights eij are nonzero only if the edges i and j are “kth order neighbors” and compose the
square N × N spatial weights matrix Ek , which represents the k-order spatial relationship. For example,
k = 1 means two road sections are directly connected; k = 2 means there is one road section between
these two road sections; k = 3 means there are two road sections between these two road sections and
so on. Then, the spatial weights matrix Ek is normalized by a line to represent the impact between road
(k)
sections, namely ∑ eij = 1. With these rules, we can construct the spatial weights matrices for the
first-order and second-order adjacency matrix. Equal weights are assigned to the kth order neighbors.
The spatial weights matrix E can be defined as the sum of the kth spatial weights matrix Ek , which is
shown in Equation (2):
K
E = ∑k=1 Ek (2)

where K is the highest order of the spatial weights matrix.

3.1.2. Measuring Spatiotemporal Similarity among road Sections

Since the current states of a road section have predictive power on the future conditions of
adjacent roads, the first step is to measure the spatiotemporal correlations among road sections.
Several approaches, including the Manhattan distance, the Euclidean distance and so forth, are suitable
to calculating the correlation [46]. Allowing for the condition that the similarities are measured under
a given time lag, the correlation distance, which is derived from Pearson’s correlation, is adopted as
a potential method of measuring the dependence of the two time series. Suppose that there are two
finite and equidistant series of X and Y; the correlation distance between these two series is given by
Equation (3):
D ( X, Y ) = 1 − Cor ( X, Y ) (3)

where Cor ( X, Y ) is Pearson’s correlation between the series X and Y. Cor ( X, Y ) is calculated by
Equation (4):
∑( X − µ X )(Y − µY )
Cor ( X, Y ) = (4)
σX σY
where µ X and µY are the mean values of the series X and Y, respectively and σX and σY are the
variances of the series X and Y, respectively.
The formulas listed above represent the basic expression of the correlation distance.
To comprehensively consider the spatial correlation of the road network with the temporal correlation,
a spatial weights matrix is introduced as a spatial indicator to improve the initial correlation
Sensors 2018, 18, 2287 6 of 26

distance. Let xit represent the traffic speed in the road section i at time t, where xit = xi1 , xi2 , . . . , xiT .

Then, the integrated speed values of the adjacent road sections can be defined as RXit , which is
calculated by Equation (5):
R
RXit = ∑ j=1 eij x tj (5)

where i ∈ [1, R] is the number of the road sections, t ∈ [1, T] is the time interval and eij is the spatial
weight from spatial weight matrix E.
Then, the correlation between the speeds of road section i and its adjacent road sections can be
calculated by Equation (6):

∑tT=1 xit − xi RXit+s − RXi

Cori (s) = q
2 q T (6)
∑tT=1 xit − xi ∑t=1 ( RXit+s − RXi )

where s = (1, 2, . . . , S), (s > 0) is the time lag between the speeds of road section i and its adjacent
road sections, xi is the mean value of the traffic speed xit in road section i during a time duration of T,
RXi is the mean value of the integrated speed RXit and the other variables are similarly defined as in
Equation (5).
Cori (s) measures the impact that the traffic conditions of road section i exert on the future state
of its adjacent road sections. A greater value of Cori (s) correlates to a greater influence, whereas the
correlation distance Di (s) is the inverse (Equation (7)):

Di (s) = 1 − Cori (s) (7)

where Di (s) stands for the correlation distance between road section i and its adjacent road sections
under a time delay of s.

3.1.3. Recognizing and Extracting Critical Road Sections

The correlation distance provides a criterion to measure the spatiotemporal similarity among
road sections. In this paper, if the correlation distance of a road section is smaller than that of the other
sections, we call it a critical road section, which means that it has a powerful impact on the future traffic
conditions of its adjacent road sections. To extract critical road sections of the traffic network, we need
to rank the forces of all road sections based on correlation distances under every different delay.
In the area of multiattribute decision analysis (MADA), several methods are commonly used to
comprehensively consider the trade-offs between multiple indicators, including the preference ranking
organization method for enrichment evaluation (PROMETHEE), elimination and choice expressing
reality (ELECTRE), analytic hierarchy process (AHP), order of preference by similarity to ideal solution
(TOPSIS) and so forth. [47]. Among all these methods, TOPSIS has been effectively used for solving
many different selection/ranking problems since it is a comprehensible, intuitive tool and easy to
perform. Applying TOPSIS to rank the impacts of road sections can integrate correlation distances
under different time lags into a rational composite index, which is called the integrated correlation
distance. The details are as follows.

• Step 1: Create an evaluation matrix.

As mentioned above, we shall regard the correlation distance matrix (CDM) with elements Di (s)
as the evaluation matrix. The CDM consists of R alternatives and S attributes that represent R road
sections and S time lags respectively; hence, CMD = ( Di (s))S× R .

• Step 2: Calculate the weights under every time lag.

Sensors 2018, 18, 2287 7 of 26

Different weights are assigned to different time lags with the consideration that closer time
intervals have more similar traffic states. Thus, the weights under time lag s are calculated using the
Euclidean distance as shown in Equation (8):
q
2
∑tT=−1s ∑iR=1 xit+s − xit
Ed(s) = (8)
T−s

where xit is the traffic speed of road section i at time t and the other variables are similar to that of
Equation (6). A smaller value of Ed(s) means the traffic states are more similar. However, a larger
weight should be assigned to a more similar traffic state; therefore, the Ed(s) is reversed and scaled to
a range of 0 to 1 by Equation (9):

Ed(s) − min( Ed)

Ed0 (s) = 1 − (9)
max ( Ed) − min( Ed)

where min(Ed) and max(Ed) denote the minimum and maximum of the Euclidean distance
set, respectively.

• Step 3: Determine the ideal worst solution A− and the ideal best solution A+ .

The correlation distance is varied with the time lag; therefore, A− and A+ need to change over
the time lag and they are defined as Equations (10) and (11), respectively:

A− (s) = {minDi (s)|s ∈ (1, 2, . . . , S), 1 ≤ i ≤ R} (10)

A+ (s) = {maxDi (s)|s ∈ (1, 2, . . . , S), 1 ≤ i ≤ R} (11)

where A− (s) and A+ (s) are the ideal worst and best solution, respectively, under time lag s and the
other variables are similar to that of Equations (7)–(9).

• Step 4: Calculate the distance between Di (s) and the ideal worst and best solutions.

The weighted Euclidean distance is applied to calculate the distance between Di (s) and the ideal
worst and the best solution. The weights are involved to identify different degrees of impact under
each time lag, which can be determined by Equations (8) and (9). The distance between road section
i and the worst solution A− (s) is calculated by Equation (12):
q
l
Di− = ∑s=1 Ed0 (s)·( Di (s) − A− (s))2 , i = 1, 2, 3, . . . , R (12)

The distance between the correlation distance of road section i and the worst solution A(s)+ is
determined by Equation (13):
q
l
Di+ = ∑s=1 Ed0 (s)·( Di (s) − A+ (s))2 , i = 1, 2, 3, . . . , R (13)

where Di− and Di+ are the weighted Euclidean distances between the correlation distance of road
section i and the worst and best solution, respectively. Other variables are similarly defined as in
Equations (7)–(11).

• Step 5: Calculate the similarity to the worst condition.

The similarity to the worst condition measures the degree of impact under all delays between
road section i and its adjacent road section, which can be calculated by Equation (14):

Di−
Ci = , 0 ≤ Ci ≤ 1, i = 1, 2, 3, . . . , R (14)
Di− + Di+
Sensors 2018, 18, 2287 8 of 26

where Ci is the similarity to the worst condition and i stands for the number of the road section,
which is the integrated correlation distance. Obviously, Ci = 1 means that the integrated correlation
distance has the best condition and Ci = 0 means that it has the worst condition.

• Step 6: Rank the impact and identify the critical road sections according to Ci (i = 1, 2, 3, . . . , R).

Ci measures the degree of impact between road section i and its adjacent road sections. A greater
value of Ci means that the traffic status of road section i has a greater impact on the future conditions
of its adjacent road sections. Thus, the impact is ordered by the value of Ci . Furthermore, we extract
a number of road sections by the percentage of α according to their orders. Then, these roads are
regard as critical road sections. α is defined as extracting rate and it actually presents the number of
critical road sections. Distinguished by the spatiotemporal features of traffic flow, the critical road
sections can be characterized and thus used to predict the traffic conditions of the overall network.
Thus, partial information can be used to predict global variables.

• Step 7: Split the time period and integrate the dataset.

Due to the unbalanced distribution of living areas and working areas, the traffic tide phenomenon
(TTP) frequently occurs in large cities [48]. To distinguish the deviation that the TTP introduces to the
spatiotemporal characteristics of the network, the change of critical road sections must be divided
into multiple periods of time in accordance with traffic pattern (i.e., peak period or ordinary period).
As a result, distinct sets of critical road sections are observed at each time period such that the number
of roads remains the same, while the elements are different. In addition, the data of every time period
are integrated into one dataset, which represents the critical road section dataset of the whole day.
The integrated dataset will be used to train the prediction model and forecast future traffic conditions
of the overall network.

3.2. Predicting the Short-Term Traffic State Using ConvLSTM NN

In this paper, the deep learning approach ConvLSTM NN is used as the prediction method.
The ConvLSTM NN presents the advantages of the CNN and LSTM methods and can simultaneously
extract spatial and temporal features.

3.2.1. Reshaping Training Data to a Spatiotemporal Matrix

Traffic information generated by the floating car with a GPS device consists of both spatial and
temporal features, which should work in conjunction to predict future traffic states of the network.
Assume that xit represents the average speed on road section i at time t. Then, the spatiotemporal matrix
is established with the elements of xit , which represents the average speed on road section i at time t.
In addition, the row and column of the matrix are spatially and temporally associated, respectively.
As a result, the daily traffic states can be mathematically transformed into a spatiotemporal matrix
with T rows and R columns, which can be denoted as Equation (15):
 
x11 , x12 , ··· x1R

 x21, x22 , ··· x2R


M= .. .. ..  (15)
. . .
 
 
x T1 , x T2 , ··· x TR

where T is the length of time intervals and R is the number of road sections.
Along the time dimension (i.e., columns of the matrix), the time range that usually extends
from the beginning to the end of the day is divided into narrow intervals of 2–15 min.
Additionally, the speeds are also aggregated and averaged over 2–15 min in coordination with the time
intervals. The length of the interval is determined primarily based on the sampling intervals and the
Sensors 2018, 18, 2287 9 of 26

data update frequency. Additionally, for real-time forecasting that focuses on the traffic evolution from
15 to 40 min, intervals that are too narrow or wide are inappropriate in practice.
Along the spatial dimension (i.e., rows of the matrix), the speeds of each road section are arranged
as a sequence representing the traffic state of the network during a certain time period. Notably, for our
Sensors 2018, 18, x FOR PEER REVIEW 9 of 24
model, the spatiotemporal matrix of critical road sections and the overall network have the same
temporal
every dimension
layer retrievesbut a different
a distinct spatial
feature dimension
using differentsince thewhile
filters critical road
the sections
latter assistsare
inextracted
reducingfromthe
the latter in the light of the spatiotemporal features. Thus, to ensure the consistency
number of parameters required for training the CNN under the premise of ensuring prediction of the two types
of matrices
accuracy. in space,
Because thethe sequence
input of the
for a CNN roads
can will not beregarded
be intuitively disrupted. In simple
as an image interms,
which weeach
draw the
pixel
speeds of critical road sections from the original matrix and then merge these columns
value is associated with one type of traffic state during a certain time, a 2D CNN is naturally utilized together into
a matrix
to abstractof spatial
critical road sections.
features Figure
between road1 displays
sections. two heat2plots
Figure of the traffic
illustrates speed. The
the structure of aprocess
CNN,
simply illustrates
including the input how to generate
layer, the spatiotemporal
convolution matrixfully
layer, pooling layer, of critical road sections
connected layer and tooutput
form aslayer.
large
an overall network matrix as possible to ensure the spatial characteristics.
Each part plays a unique and vital role for a CNN and the details are briefly explained below.

Sensors 2018, 18, x FOR PEER REVIEW 9 of 24

every layer retrieves a distinct feature using different filters while the latter assists in reducing the
number of parameters required for training the CNN under the premise of ensuring prediction
accuracy. Because the input for a CNN can be intuitively regarded as an image in which each pixel
value is associated with one type of traffic state during a certain time, a 2D CNN is naturally utilized
to abstract spatial features between road sections. Figure 2 illustrates the structure of a CNN,
including the input layer, convolution layer, pooling layer, fully connected layer and output layer.
Each part plays a unique and vital role for a CNN and the details are briefly explained below.

Figure
Figure 1.
1. The
The process
process of
of shaping
shaping spatiotemporal
spatiotemporal matrix
matrix of
of critical
critical road
road section:
section: (a)
(a) A
A spatiotemporal
spatiotemporal
matrix
matrix of
of overall
overall network
network and
and (b)
(b) AA spatiotemporal
spatiotemporal matrix
matrix of
of critical
critical road sections.

3.2.2. Capturing Spatial Features via a CNN

The CNN has been successfully applied to traffic predictions because of its great potential to
extract features using multiple layers. A typical CNN mainly consists of multiple convolution layers
and pooling layers. The former contributes to mining spatial dependencies of road sections since every
layer retrieves a distinct feature using different filters while the latter assists in reducing the number
of parameters required for training the CNN under the premise of ensuring prediction accuracy.
Because the input for a CNN can be intuitively regarded as an image in which each pixel value is
Figure 2. The structure of the convolutional neural networks (CNN).
associated with one type of traffic state during a certain time, a 2D CNN is naturally utilized to abstract
spatial features
Suppose between
that we needroad sections.the
to predict Figure
future2 illustrates theof
traffic speed structure of a CNN,
the network 𝑉 𝑡+𝑎 including the
= {𝑥𝑖𝑡+𝑎 }𝑅𝑖=1 input
, where
alayer,
is theconvolution
Figure
prediction layer, pooling
1. The process
horizonofand R layer,
shaping fully connected
spatiotemporal
is the number layer
matrix
of road and The
output
of critical
sections. road layer.
section:
input Each partisplays
(a) CNN
of the A spatiotemporala unique
the historical
and vital role
matrix of for a CNN
overall andand
network the(b)
details
A are briefly
𝑡−𝑛 𝑡−1
spatiotemporal explained below.
𝑡−𝑛 road 𝑡−𝑛
𝑡matrix of critical 𝑃𝛼
sections.
traffic speed of critical road sections {𝑈 , 𝑈 , 𝑈 }, where 𝑈 = {𝑥𝑖 }𝑖=1 represents the traffic
state of critical road sections at time (t−n), n is the look back step and 𝑃𝛼 is the number of critical road
sections. Then, the spatial features of input are captured by convolutional and pooling layers. Let 𝑂𝑟𝑙
be the output of lth convolutional and pooling layers with r filters and the weights and bias of lth
layers be (𝑊𝑟𝑙 , 𝑏𝑟𝑙 ). Then, the 𝑂𝑟𝑙 can be calculated by Equation (16):

𝑂𝑟𝑙 = 𝑝𝑜𝑜𝑙(𝑓(∑ 𝑊𝑟𝑙 ∗ 𝑂𝑟𝑙−1 + 𝑏𝑟𝑙 )) (16)

𝑟

where 𝑂𝑟𝑙−1 represents the output of the previous layer, in which 𝑂𝑟1 represents the input layer; f is
a nonlinear activation function;
Figure
Figure 2. The
2. and pool of
Thestructure
structure denotes
ofthe the pooling
theconvolutional
convolutional procedure.
neural
neural networks (CNN).
networks (CNN).

3.2.3.Suppose
Capturing thatTemporal
we need Features
to predictVia
thethe LSTM
future traffic speed of the network 𝑉 𝑡+𝑎 = {𝑥𝑖𝑡+𝑎 }𝑅𝑖=1 , where
a is the predictiontraffic
Intuitively, horizon andat
states R is the moment
each number of roada sections.
have The input
strict sequential of the CNNinis the
relationship the historical
temporal
𝑡−𝑛 𝑡−1 𝑡 𝑡−𝑛 𝑡−𝑛 𝑃𝛼
traffic speed of critical road sections {𝑈 , 𝑈 , 𝑈 }, where
dimension rather than being isolated from each other, which is especially 𝑈 = {𝑥𝑖 } represents the
𝑖=1 suitable for RNNs traffic
to
state of critical road sections at time (t−n), n is the look back
capture the temporal evolution of traffic flow. However, it is difficult forstep and 𝑃 is the number of critical
𝛼 traditional RNNs to capture road
sections. Then, the spatial features of input are captured by convolutional and pooling layers. Let 𝑂𝑙
Sensors 2018, 18, 2287 10 of 26

R
Suppose that we need to predict the future traffic speed of the network V t+a = xit+ a i=1 ,


where a is the prediction horizon and R is the number of road sections. The input of the CNN is the
P
historical traffic speed of critical road sections {U t−n , U t−1 , U t }, where U t−n = xit−n i=α 1 represents


the traffic state of critical road sections at time (t−n), n is the look back step and Pα is the number of
critical road sections. Then, the spatial features of input are captured by convolutional and pooling
layers. Let Orl be the output of lth convolutional and pooling layers with r filters and the weights and
bias of lth layers be (Wrl , brl ). Then, the Orl can be calculated by Equation (16):
  
Orl = pool f ∑r Wrl ∗ Orl −1 + brl (16)

where Orl −1 represents the output of the previous layer, in which Or1 represents the input layer; f is
a nonlinear activation function; and pool denotes the pooling procedure.

3.2.3. Capturing Temporal Features Via the LSTM

Intuitively, traffic states at each moment have a strict sequential relationship in the temporal
dimension rather than being isolated from each other, which is especially suitable for RNNs to capture
the temporal evolution of traffic flow. However, it is difficult for traditional RNNs to capture the
temporal
Sensors 2018,dependency
18, x FOR PEERifREVIEW
two time intervals are remote. Thus, an LSTM, which is a specific form
10 of of
24
a RNN, is proposed to tackle these issues by adding memory cells in hidden layers. As shown in
Figure 3, four main parts (input gate, neuron with a self-recurrent connection, forget gate and output
gate) collaborate to alleviate the problems of traditional RNNs caused by the gradient vanish and
problems.
explosion problems.

Figure 3. The architecture of long short-term memory neural network (LSTM

(LSTM NN).
NN).

In our
our model,
model,nextnexttotothetheCNN,
CNN,the theLSTM
LSTM naturally takes 𝑉 𝑡= R𝑡 }𝑅
t {𝑥
In naturally takes thethe output
output of 𝑞
of the
the CNN CNN
V t = x i
𝑖 𝑖=1
as
as its input to predict the future traffic states, namely, its output 𝐻 = {ℎt 𝑡 }q , where q is the number i =1
𝑡
its input to predict the future traffic states, namely, its output Ht = hi i=1 , where q is the number 𝑖 𝑖=1
of hidden
of hidden units
units of
of the
the output
output layer.
layer. For
For memory
memory cell, cell, the
theinput
inputstate stateisis 𝐺G𝑡−1 while the output is 𝐺𝑡 .
t−1 while the output is Gt .
Meanwhile, the
Meanwhile, the states
states ofof input,
input, forget
forget and
and output
output gates
gates areare I𝐼t𝑡,, F𝐹t𝑡 and
and O 𝑂𝑡 ,, respectively. Clearly, due to
t respectively. Clearly, due to
some relatively simple linear processes (i.e., plus or scalar product),
some relatively simple linear processes (i.e., plus or scalar product), a long-term state can be a long-term state can be reserved
reserved
by
by Gt−𝐺 . However,
𝑡−1. However, H 𝑡−1 𝐻 represents
represents the the
state state
of lastof last
memory memory
cell. cell.
Through Through
the the comprehensive
comprehensive operation
1 t −1
operation
of the threeofgates,
the three
long-termgates,andlong-term and current
current states states
exert effect onexert futureeffect
memory on future memory cells
cells simultaneously.
simultaneously. The temporal features can be iteratively
The temporal features can be iteratively calculated by Equations (17)–(22): calculated by Equations (17)–(22):
Input gate:  𝐼𝑡 = 𝜎(𝑊𝑣𝑖 𝑉 𝑡 + 𝑊ℎ𝑖 𝐻𝑡−1  + 𝑏𝑖 ), (17)
i t i t −1
Input gate : It = σ Wv V + W𝑓h H𝑡 +𝑓bi , (17)
Forget gate: 𝐹𝑡 = 𝜎(𝑊𝑣 𝑉 + 𝑊ℎ 𝐻 𝑡−1 + 𝑏𝑓 ), (18)

Output gate: 𝑂f 𝑡 t−1 𝑜 𝑡−1 (19)

 
Forget gate : Ft = σ W 𝑂𝑡vf =
V t𝜎(𝑊
+W 𝑣 h𝑉H + 𝑊 +ℎ 𝐻b f , + 𝑏𝑜 ), (18)
Cell input:  = 𝑡𝑎𝑛ℎ(𝑊 𝑐 𝑉 𝑡 + 𝑊 𝑐 𝐻
𝐺𝑡−1 𝑡−1 + 𝑏 ), (20)
𝑣 ℎ 𝑐
Output gate : Ot = σ WvO V t + Who H t−1 + bo , (19)
Cell output: 𝐺𝑡 = 𝐼𝑡 ⊙ 𝐺𝑡 + 𝐹𝑡 ⊙ 𝐺𝑡−1 , (21)
Hidden layer output: 𝐻𝑡 = 𝑂𝑡 ⊙ tanh(𝐺𝑡 ), (22)

where the weight matrix 𝑊 and bias vector b are constructed to connect the input layer, output layer
and memory cell, ⊙ denotes the scalar product of two vectors and 𝜎(∙) represents the standard
logistics sigmoid function defined in Equation (23):
1
σ(𝑥) = (23)
1 + 𝑒 −𝑥
Sensors 2018, 18, 2287 11 of 26

 
Cell input : Gt−1 = tanh Wvc V t + Whc H t−1 + bc , (20)

Cell output : Gt = It  Gt + Ft  Gt−1 , (21)

Hidden layer output : Ht = Ot  tan h( Gt ), (22)

where the weight matrix W and bias vector b are constructed to connect the input layer, output layer
and memory cell,  denotes the scalar product of two vectors and σ (·) represents the standard logistics
sigmoid function defined in Equation (23):

1
σ( x ) = (23)
1 + e− x

3.2.4. Training with the ConvLSTM NN

Integrated with the advantages of the CNN and LSTM methods, the ConvLSTM NN is utilized
to predict future traffic states by sufficiently exploiting the spatiotemporal characteristics of the data.
Eventually, a fully connected layer is employed to predict the future speed by taking the output of the
LSTM as the input. The future speed can be calculated by Equation (24):

Y t+1 = Wy Ht + by (24)

where Wy and by are the weight and bias related to the hidden layer. Conclusively, the model is trained
from end to end and the values of Y t+1 are prediction results that represent the output of the entire
mode. Several hyperparameters within the model will be set and elaborated in the experiment section.
Notably, the input size will vary as the number of critical road sections changes due to different
extracting rates α; hence, several other hyperparameters will also change.

4. Empirical Study in Beijing Network

4.1. Description of Data Used

To valid the efficiency of our methodology, data collected by a taxi equipped with GPS devices
from 1 June 2015 to 13 October 2015 (92 working days) in Beijing, the capital of China, are utilized
to train the ConvLSTM NN model and predict the future traffic speed of network. The updating
frequency of the data is 2 min and the time period from 6:00:00 to 23:00:00 is considered for the high
travel demand period, which has been repeatedly observed. Accounting for the condition that the
traffic state varies with every time interval, we can observe 511 traffic states per day.
In our paper, a sub-transportation network near West Second Ring Road is selected as the research
objective, as shown in Figure 4. The network consists of 278 road sections, including several types of
road hierarchies, such as freeways, arteries, secondary roads and collectors. The total length of all the
roads is approximately 24.53 km and the network covers an approximate 0.6-km2 area. Two thirds of
the data (61 days) are regarded as training data, while the remainder (31 days) compose the testing
data. Additionally, the structural missing data marked by pentagram can be obviously observed.
 traffic state varies with every time interval, we can observe 511 traffic states per day.
In our paper, a sub-transportation network near West Second Ring Road is selected as the
research objective, as shown in Figure 4. The network consists of 278 road sections, including several
types of road hierarchies, such as freeways, arteries, secondary roads and collectors. The total length
of all the roads is approximately 24.53 km and the network covers an approximate 0.6-km2 area. Two
thirds of the data (61 days) are regarded as training data, while the remainder (31 days) compose the
Sensors 2018,testing
18, 2287data. Additionally, the structural missing data marked by pentagram can be obviously 12 of 26
observed.

Figure 4. Layout of the sub-transportation network in Beijing.

Figure 4. Layout of the sub-transportation network in Beijing.
4.2. Critical Road Sections Identification by STCA
4.2. Critical Road
TheSections Identification
spatial similarity byflow
of traffic STCAcanbe easily obtained according to the topology of the
network, which will be viewed as the static properties of the traffic flow. The k-order spatial weights
The spatial similarity of traffic flow can be easily obtained according to the topology of the
matrix is constructed to describe the topology of the traffic network. The spatial influence is measured
network, which will be
by calculating theviewed
Euclideanasdistance
the static properties
between different of
roadthe trafficasflow.
sections shownThe k-order(25):
in Equation spatial weights
matrix is constructed to describe the topology√∑of𝑇 the
𝑡 traffic
𝑡 network. The spatial influence is measured
𝑡=1(𝑥𝑖 − 𝑥𝑖+𝑘 )) (25)
𝑆𝑃(𝑘) =
by calculating the Euclidean distance between different road sections as shown in Equation (25):
𝑇
where 𝑆𝑃(𝑘) is the Euclidean distance that r measuresthe spatial correlation between roads, k denotes
k-order, T is the total time period and 𝑥𝑖𝑡 represents

∑tT=1 thexitspeed
− xit+ofk road
) section i at time t.
Figure 5 depicts obviously positive correlation between the Euclidean distance and the k value
SP(k) = (25)
when k is less than 7. It is obvious that if k is bigger than
T 7, distant roads have weak impact on the
given road. And from Figure 5, we can find that the variation between 5th and 6th order is less than
where SP(previous
k) is thevalues.
Euclidean distance
Therefore, we takethat measures
5-order theasspatial
neighbors adjacent correlation between
road sections. roads,
Based on these k denotes
k-order, T is the total time period and xit represents the speed of road section i at time t.
Figure 5 depicts obviously positive correlation between the Euclidean distance and the k value
when k is less than 7. It is obvious that if k is bigger than 7, distant roads have weak impact on the
given road. And from Figure 5, we can find that the variation between 5th and 6th order is less than
previous values. Therefore,
Sensors 2018, we
18, x FOR PEER take 5-order neighbors as adjacent road sections. Based on
REVIEW 12 ofthese
24 results,
we can conclude that only the nearby roads contribute to the current road section traffic speed while
results, we can conclude that only the nearby roads contribute to the current road section traffic speed
the remotewhile
onesthe have little influence. Moreover, a sensitivity analysis is also conducted to measure the
remote ones have little influence. Moreover, a sensitivity analysis is also conducted to
impact of measure
different spatial
the impact of weights
differentmatrices to the
spatial weights predictive
matrices performance.
to the predictive performance.

Figure 5. Influence between roads in the case of different k-order.

Figure 5. Influence between roads in the case of different k-order.
To capture the stochastic and time-varying dynamic characteristics, the spatiotemporal features
of the traffic flow should be mined by the spatial and temporal correlations. Considering the traffic
flow is commonly repeated every working day [49], we average the speeds of all 92 days, which
means an entry of the average speed matrix represents the average speed along 92 days in a certain
road section at a moment. It can be seen that the mean of the speed is gradually tending to
convergence with the increasing of the days. For most roads, the mean of the speeds becomes steady
Sensors 2018, 18, 2287 13 of 26

To capture the stochastic and time-varying dynamic characteristics, the spatiotemporal features of
Sensors 2018, 18, x FOR PEER REVIEW 13 of 24
the traffic flow should be mined by the spatial and temporal correlations. Considering the traffic flow
is commonly repeated every working day [49], we average the speeds of all 92 days, which means an
entry of the average speed matrix represents ( Di ( s))the , for speed
average
120278 MPP along 92 days in a certain road section
( D (of
at a moment. It can be seen that the mean  the speed is gradually
s)) 210278 , for DOP tending to convergence with the
CMD   i
increasing of the days. For most roads, the mean of the speeds becomes steady around 60 days. We(26) take
( Di ( s))120278 , for EPP
the average of 92 days to make sure as(much as speeds becomes steady. In this way, we can possibly
 Di ( s))61278 , for EOP
avoid special situations and extract the most universal critical road sections. Then the spatiotemporal
where
features278
canisbe
theextracted
total number of the
from the road speed
average sections. Then,
matrix. Asthe weights used
previously to evaluate
discussed, the impact
similarities between of
time
road lags ranging
sections from 1 to by
are measured 30 integrated
time steps correlation
are calculated by Equations
distances (8) and (9).
under different Figure
time lags. 7Figure
clearly6
shows that the interaction correlation
the spatiotemporal between two traffic states
distances weakens
of several road as the minute
sections whenincreases, which
time ranges fromis2why
min
we pay
to 60 more
min, attention
which to proximate
represents traffic
1 to 30 time states
delays when ranking the influence of roads.
respectively.

Figure
Figure 6.
6. Trend
Trend of
of correlation
correlation distance
distance along
along with the time
with the time lag.
lag.

Figure 6 shows that the spatiotemporal correlation distance values increase as the time lag extends,
which means that the states of the adjacent moments are more similar than the distant moments.
In other words, the traffic states of one road have a stronger predictive power for the state of the nearby
moment of the neighboring road network than for the distant moment. In addition, the curves for
different roads present distinct growth trends, which characterize various temporal features.
Another important factor that should be taken into account for such predictions is the influence
of TTP. The temporal features of traffic flow vary dynamically over time, which leads to the transition
of the critical road sections. For example, one road section may play a relatively considerable role
in the morning peak time rather than the evening peak time. By analyzing the evolution of traffic
flow and using previous empirical knowledge [50], one day is divided into four periods in accordance
with different traffic patterns: the morning peak period (6:00:00–10:00:00) (MPP), the daytime ordinary
period (10:00:00–17:00:00) (DOP), the evening peak period (17:00:00–21:00:00) (EPP) and the evening
ordinary period (21:00:00–23:00:00)
Figure 7. (EOP).
Trend of weights along with the time lag.
Subsequently, TOPSIS is employed to measure the degree of predictive power among road sections
duringTheeach
ideal worst
time andand
period bestthen
solutions
extractare the worst
critical and the best
road sections valuestoofthe
according theintegrated
correlation distance
correlation
under eachFirst,
distances. timethe
lag,evaluation
respectively. In addition,
matrix the integrated
of each period correlation
is constructed. length𝐶of
distance
Since the 𝑖 that
eachmeasures
period is
the influence of the road on its adjacent road sections is calculated by the weighted
different, the size of the CDM varies based on the context of each period as shown in Equation (26): Euclidean
distance. When correlation distances  corresponding to all time lags are taken into account, the
spatiotemporal characteristics of the road  (sections


Di (s))120will
×278be f or MPP
, reflected comprehensively.
Eventually, the impact of roads is (ranked
 D (s)) , f or DOP
i according
210×278 to the integrated correlation distance.
CMD = (26)
Notably, to measure the performance under  ( Didifferent
 (s))120×278 or EPP rates 𝛼 (i.e., the proportion of
, fextracting
road

critical road sections to all roads), the number

( Di (of
s))critical f orsections
61×278 ,road EOP that are input to the prediction
model varies. Then, the number of critical road sections corresponding to 𝛼 is set to 𝛽 . The
correspondence of the extracting rate and the number of critical road sections is listed in Table 1. For
example, α = 0.70 means that we will take 195 roads as critical road sections and subsequently use
them to predict the traffic states of 278 roads. Figure 8 shows the degree of impact of all roads at
different time periods. The red line indicates that the road section has a greater influence on the future
Sensors 2018, 18, 2287 14 of 26

where 278 is the total number of the road sections. Then, the weights used to evaluate the impact
of time lags ranging from 1 to 30 time steps are calculated by Equations (8) and (9). Figure 7 clearly
shows that the interaction between two traffic states weakens as the minute increases, which is why
we pay more attentionFigure
to proximate traffic
6. Trend of states distance
correlation when ranking the the
along with influence of roads.
time lag.

Figure
Figure 7.
7. Trend
Trend of
of weights
weights along
along with
with the
the time
time lag.
lag.

The ideal worst and best solutions are the worst and the best values of the correlation distance
The ideal worst and best solutions are the worst and the best values of the correlation distance
under each time lag, respectively. In addition, the integrated correlation distance 𝐶𝑖 that measures
under each time lag, respectively. In addition, the integrated correlation distance Ci that measures the
the influence of the road on its adjacent road sections is calculated by the weighted Euclidean
influence of the road on its adjacent road sections is calculated by the weighted Euclidean distance.
distance. When correlation distances corresponding to all time lags are taken into account, the
When correlation distances corresponding to all time lags are taken into account, the spatiotemporal
spatiotemporal characteristics of the road sections will be reflected comprehensively.
characteristics of the road sections will be reflected comprehensively.
Eventually, the impact of roads is ranked according to the integrated correlation distance.
Eventually, the impact of roads is ranked according to the integrated correlation distance.
Notably, to measure the performance under different road extracting rates 𝛼 (i.e., the proportion of
Notably, to measure the performance under different road extracting rates α (i.e., the proportion
critical road sections to all roads), the number of critical road sections that are input to the prediction
of critical road sections to all roads), the number of critical road sections that are input to the
model varies. Then, the number of critical road sections corresponding to 𝛼 is set to 𝛽 . The
prediction model varies. Then, the number of critical road sections corresponding to α is set to
correspondence of the extracting rate and the number of critical road sections is listed in Table 1. For
β. The correspondence of the extracting rate and the number of critical road sections is listed in Table 1.
example, α = 0.70 means that we will take 195 roads as critical road sections and subsequently use
For example, α = 0.70 means that we will take 195 roads as critical road sections and subsequently
them to predict the traffic states of 278 roads. Figure 8 shows the degree of impact of all roads at
use them to predict the traffic states of 278 roads. Figure 8 shows the degree of impact of all roads at
different time periods. The red line indicates that the road section has a greater influence on the future
different time periods. The red line indicates that the road section has a greater influence on the future
traffic state of adjacent roads in the MPP, DOP, EPP or EOP. The distribution of the red line indicates
that the critical road sections vary with time, which verifies the effect of the TTP on the critical degree
of the road sections. From Figure 8, we can find that roads on the left side (i.e., the West Second Ring
Road) appears to be critical all the periods. The reason is that these roads are expressways which
undertake major traffic flow in the network. Another interesting fact is that roads in the central of
the network has more critical degrees during peak hours than ordinary hours. It is probably because
that compared with expressway (i.e., the West Second Ring Road), urban roads are affected by traffic
signals and they show more various spatiotemporal features. Therefore, more roads are needed to
predict the traffic states of nearby road sections.

Table 1. Correspondence of extracting rate and the number of critical road sections.

Parameters Values
Extracting rate α 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Number of roads β 167 181 195 209 222 236 250 264 278
 Table 1 Correspondence of extracting rate and the number of critical road sections.

Parameters Values
Extracting
Sensors 2018, 18, 2287
rate α 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
15 of 26
Number of roads 𝛽 167 181 195 209 222 236 250 264 278

Figure 8. Distribution of critical road sections in the: (a) Morning Peak Period; (b) Daytime Ordinary
Figure 8. Distribution of critical road sections in the: (a) Morning Peak Period; (b) Daytime Ordinary
Period; (c) Evening Peak Period; and (d) Evening Ordinary Period.
Period; (c) Evening Peak Period; and (d) Evening Ordinary Period.

4.3. Training with Critical Road Sections based on the ConvLSTM NN

4.3. Training with Critical Road Sections based on the ConvLSTM NN
4.3.1. Generating
4.3.1. Generating Input
Input Data
Data
In aa traffic
In trafficnetwork,
network, nearby
nearby road
road sections
sections are
are more
morerelative
relative that
that distant
distant ones.
ones. In
In order
order to
to retain
retain
the spatial
the spatial correlation
correlation between
betweenadjacent
adjacentroads,
roads,it itis is
necessary
necessary to to
arrange
arrangeadjacent roads
adjacent together.
roads An
together.
effective method is to straighten consecutive roads into a line and then construct
An effective method is to straighten consecutive roads into a line and then construct a spatiotemporal a spatiotemporal
state matrix.
state matrix. As As stated
statedininMa
Maetetal.al.(2017) [36],
(2017) a simple
[36], a simple traffic network,
traffic network,such as an
such asexpressway
an expresswayor ring
or
road, can be easily straightened to generate a straightforward matrix. However,
ring road, can be easily straightened to generate a straightforward matrix. However, complete spatial complete spatial
dependencies cannot be retained when reshaping the road sections of the complicated network into
a straight line. To alleviate the negative effect caused by the loss of spatial relations, the network is
segmented into successive routes and the roads are numbered along the route. Moreover, the CNN can
capture spatial features from local connections despite the complex and network-wide dependencies
of traffic speeds. After reshaping the speed of all road sections into a spatiotemporal matrix, we obtain
92 matrices and each represents the daily traffic state and are regarded as the labels of the prediction
model. Using the same approach, spatiotemporal matrices that denote the traffic states of critical road
sections can be generated and regarded as the input of the model.

4.3.2. Tuning up with the ConvLSTM

Without a general regulation for constructing the structure of DL NNs, the selection of
hyperparameters mainly depends on expert experience. In addition, compared with the prevalent
CNN or LSTM NN, the spatial dimensions of the inputs and outputs in our CRS-ConvLSTM are
Sensors 2018, 18, 2287 16 of 26

unequal. In this section, we take 195 critical road sections (i.e., extracting rate α = 0.70) as an example
to explain the procedure for constructing the prediction model.
In our model, the length of the time steps is set to 15, which means that 30 min of historical
data are utilized to predict the traffic state of the next 2 min. For a 2D CNN structure, a 5-layer
fully convoluted structure is used so that the depth of the CNN is neither too large nor too small.
Next, a 2-layer LSTM structure is employed to capture temporal features. The size of the convolutional
filters and max-poolings refers to the experiments of AlexNet [51] and LeNet [52], which both made
remarkable achievements in the development of CNNs. The details of the parameters are listed in
Table 2. The dimension of the input layer (1, 195, 15) is determined by the size of the input data
shape, where the first number denotes the channel of the input data is 1, the second is the number
of critical road sections at extracting rate of 0.70 and the third represents the input time horizon,
which is 15 time steps. The parameters of convolutional layer indicate the number and the size of
the filters. For example, the convolutional layer of layer 1 has 16 filters and each size of them is (3, 3).
With the size of (2, 2), the pooling layers shrink the output of each convolutional layer. The output
dimensions of CNN in layer 5 are (256, 7, 1), which are then flattened into a vector with a dimension of
1792. Before feeding the data into two layers of LSTM, the vector is transformed with a dimension of
278 through a fully-connected layer. The dimension of the output layer is 278, which represents the
total number of all the road sections.

Table 2. Hyperparameters of the hybrid convolutional long short-term memory neural network model
based on critical road sections (CRS-ConvLSTM) when α = 0.70.

Layer Name Parameters Dimensions

0 Input – (1, 195, 15)
Convolution (16, 3, 3) (16, 195, 15)
Pooling (2, 2) (16, 98, 8)
1
Activation(relu) – –
Batch-normalization – –
Convolution (32, 3, 3) (32, 98, 8)
Pooling (2, 2) (32, 49, 4)
2
Activation(relu) – –
Batch-normalization – –
Convolution (64, 3, 3) (64, 49, 4)
Pooling (2, 2) (64, 25, 2)
3
Activation(relu) – –
Batch-normalization – –
Convolution (128, 3, 3) (128, 25, 2)
Pooling (2, 2) (128, 13, 1)
4
Activation(relu) – –
Batch-normalization – –
Convolution (256, 3, 3) (256, 13, 1)
Pooling (2, 2) (256, 7, 1)
5
Activation(relu) – –
Batch-normalization – –
6 Flatten – (1792, )
7 Fully connected 278 (278, )
LSTM 200 –
8
Activation(tanh) – –
LSTM 200 –
9
Activation(tanh) – –
10 Fully connected 278 (278, )
Sensors 2018, 18, 2287 17 of 26

The ADADELTA is chosen as the optimizer because of its dynamic adaptability over time [53].
The mean squared error (MSE) is selected as the loss function and 20% of the training data are extracted
as the validation dataset. A batch-normalization layer is used to accelerate the training process [54].
Additionally, the dropout layer and early stopping are used to prevent overfitting [55].

5. Results and Comparison

5.1. Performance under Different Spatial Weight Matrices

In our paper, we propose a spatial weight matrix to capture the topology features of the network.
In the previous section (i.e., 4.2), we have qualitatively described the influence between the road and
its k-order road sections. And we can conclude that it is better to set k less than 7. Here, we adopted
root mean square error (RMSE) and root mean squared error proportional (RMSEP) to evaluate the
performance of the model in the case of different spatial weight matrices; the relevant formulas are
show in Equations (27) and (28):

1
q
N
RMSE = n0.7 ∑i=0.71 (ŷi − yi )2 (27)
n0.7
s
100 1 N
RMSEP = ave
ŷ n0.7 ∑i=0.71 (ŷi − yi )2 (28)

where ŷi is the ith ground truth value, yi is the ith predicted value and ŷ ave is the average of the
ground truth values. The value of n0.7 denotes the number of critical road sections at the extracting
rate α = 0.70 and N0.7 is the total number of traffic states.
What to be noted is that we extract 195 critical road sections (i.e., extracting rate α = 0.70) as input.
The results under different spatial weight matrices are listed in Table 3. It is obvious that the model
achieves the best when k is 5. Consequently, we set k to 5 which means we regard at most 5-order
neighbors as adjacent road sections.

Table 3. Performance in the case of different value of k.

The Value of k
Index
3 4 5 6
RMSE 6.972 6.967 6.955 6.989
RMSEP (%) 21.933 21.963 21.802 21.832

5.2. Performance between Several DL Algorithms

In order to test the performance of our CRS-ConvLSTM model, four popular deep learning-based
algorithms are selected for comparison: ANN, CNN, LSTM and SAE. The ANN adopts
a plain and shallow structure (Table A1) to process multidimensional and nonlinear problems.
Additionally, the structures of the CNN (Table A2) and LSTM (Table A3) refer to the first part and
latter part of the CRS-ConvLSTM configuration, respectively. The parameters of the SAE (Table A4)
are set according to [31]. Using the same approach as the CRS-ConvLSTM implementation, 30 min of
historic traffic speed for 195 critical road sections is used as the input to predict the overall traffic states
after 2 min. The RMSE and RMSEP are employed to evaluate the performance of all the models.
Table 4 presents the quantitative results of the CRS-ConvLSTM, ANN, CNN and LSTM and shows
that the CRS-ConvLSTM outperforms the other models, thus indicating that our CRS-ConvLSTM
model can precisely mine the spatiotemporal features of the data and generate relatively accurate
predictions. Among the rival algorithms, the LSTM has the best performance, which is likely because
the temporal features of the time-series data are essentially prominent. The results of the ANN and
SAE demonstrate that these two models fail to extract spatiotemporal characteristics that have a vital
The RMSE and RMSEP are employed to evaluate the performance of all the models.
Table 4 presents the quantitative results of the CRS-ConvLSTM, ANN, CNN and LSTM and
shows that the CRS-ConvLSTM outperforms the other models, thus indicating that our CRS-
ConvLSTM model can precisely mine the spatiotemporal features of the data and generate relatively
accurate predictions. Among the rival algorithms, the LSTM has the best performance, which is likely
Sensors 2018, 18, 2287 18 of 26
because the temporal features of the time-series data are essentially prominent. The results of the
ANN and SAE demonstrate that these two models fail to extract spatiotemporal characteristics that
have a on
impact vital impact on
prediction. prediction.
Figure Figure
9 depicts 9 depicts
the error betweenthethe
error between
ground truththe
andground truth and
the estimated the
values
estimated
under values
different under It
models. different models.
can be found It can
that be found
the error that
varies thethe
with error
timevaries
and with the time the
day, however anderror
day,
however the error of the CRS-ConvLSTM is closer
of the CRS-ConvLSTM is closer to zero most of the time. to zero most of the time.

Table 4. Performance of different prediction

prediction models when αα=
models when = 0.70.
0.70.

Model Model RMSE

RMSE RMSEP (%) RMSEP (%)
CRS-ConvLSTM 6.955 21.802
CRS-ConvLSTM 6.955 21.802
LSTM LSTM 7.359
7.359 22.849 22.849
CNN CNN 7.937
7.937 26.521 26.521
SAE SAE 8.377
8.377 26.442 26.442
ANN ANN 9.504
9.504 37.870 37.870

Figure 9.
Figure 9. Errors
Errorsbetween
betweenthe ground
the truth
ground andand
truth thethe
estimated value
estimated on: (a)
value on:3 (a)
August 2015 and
3 August 2015(b) 17
and
August 2015.
(b) 17 August 2015.

5.3. Performance
5.3. Performance under
under Different
Different Extracting
Extracting Rates
Rates
To explore
To explorethe
theeffect
effectofofthe
the extracting
extracting rate
rate α on
α on the the prediction
prediction performance,
performance, we further
we further traintrain
and
and test the model based on a different number of critical road sections according
test the model based on a different number of critical road sections according to Table 1. The RMSEs to Table 1. The
RMSEs and RMSEPs in the context of different extracting rates α are listed in Table
and RMSEPs in the context of different extracting rates α are listed in Table 5 and the results show 5 and the results
showinthat
that in the of
the range range of0.9,
0.6 to 0.6 the
to 0.9, the performance
performance of the is
of the model model is slightly
slightly superiorsuperior to the
to that of thatoverall
of the
overall prediction
prediction model
model (i.e., e. , α Here,
α =(i.1.0). = 1.0).weHere, wea define
define a drop
drop rate rate to measures
to measures theofdecline
the decline of the
the accuracy
accuracy compared with the overall prediction model as
compared with the overall prediction model as shown in Equation (29): shown in Equation (29):

Index1 − Index2
drop rate = (29)
Index1
Sensors 2018, 18, x FOR PEER REVIEW 18 of 24

𝐼𝑛𝑑𝑒𝑥1 − 𝐼𝑛𝑑𝑒𝑥2
Sensors 2018, 18, 2287 drop rate = 19 of 26
(29)
𝐼𝑛𝑑𝑒𝑥1
For drop rate in Table 5, 𝐼𝑛𝑑𝑒𝑥1 denotes the error index (i.e., RMSE or RMSEP) at the case of the
Table 5. Performance of the CRS-ConvLSTM under different extracting rate.
overall prediction model and 𝐼𝑛𝑑𝑒𝑥2 presents the RMSE or RMSEP at the extracting rate of 𝛼.
The decline of the accuracy compared with the overall prediction model is ranging from 1.638%
Prediction Performance on Test Dataset
to 10.180% for the ff RMSE and 1.841% to 14.537% for the RMSEP. In general, the decrease in accuracy
RMSE Drop Rate (%) RMSEP (%) Drop Rate (%)
is reasonable and within the acceptable limits, which demonstrates the validity and generalizability
of the approach and 1.0 indicates6.778
that certain road- sections account 21.022 -
for a lower contribution toward the
0.95 6.889 1.638 21.530 2.417
prediction. Among all
0.90
the extracting
7.144
rates at less than
5.400
1.0, the value
22.698
of 0.95 achieves
7.973
the optimal result,
which is primarily 0.85because the 7.145most minor5.415 road sections22.693are removed. Additionally,
7.949 when the
extracting rate reaches
0.80 a value
7.025of 0.65, the3.644
predictive performance
21.425 gradually
1.917 tends to become
unstable, which is 0.75 6.926 too many2.184
likely because roads are omitted.21.409Figure 10 shows
1.841 the trend of the
0.70 6.955 2.611 21.802 3.710
prediction accuracy, which generally declines as the extracting rate decreases. It is interesting that we
0.65 7.232 6.698 23.763 13.039
can discovery a 0.60 bump up around 7.468 0.85, which 10.180can also be observed in [56].They
24.078 14.537 utilized SVD-
combined tensor decomposition to remedy incomplete traffic data. When filling up the fiber-like
missing data (i.e., structural missing data), the RMSE and the missing rate is not a simple linear
For drop rate in Table 5, Index1 denotes the error index (i.e., RMSE or RMSEP) at the case of the
relationship and a valley is presented around 40%. Specifically, the RMSE decreases with the increase
overall prediction model and Index2 presents the RMSE or RMSEP at the extracting rate of α.
of missing rate at the first. However, when missing rate is beyond 40%, the RMSE increases with the
The decline of the accuracy compared with the overall prediction model is ranging from 1.638% to
rise of missing rate. The same trend can be witnessed when extracting rate is lower than 0.85 in our
10.180% for the RMSE and 1.841% to 14.537% for the RMSEP. In general, the decrease in accuracy is
paper. One of the reasons is that recovering structural missing data is much more difficult than those
reasonable and within the acceptable limits, which demonstrates the validity and generalizability of
of random missing [57]. It is also probably because of the distribution of critical road sections.
the approach and indicates that certain road sections account for a lower contribution toward the
prediction. Among Tableall the extracting
5. Performance rates
of the at less than under
CRS-ConvLSTM 1.0, the value extracting
different of 0.95 achieves
rate. the optimal
result, which is primarily because the most minor road sections are removed. Additionally, when the
extracting rateRate
reaches Prediction Performance on Test Dataset
Extracting 𝛂 a value of 0.65, the predictive performance gradually tends to become unstable,
which is likely because too many RMSE roads are Drop RateFigure
omitted. (%) 10 RMSEP
shows the (%)trend Drop
of theRate (%)
prediction
1.0 6.778 - 21.022
accuracy, which generally declines as the extracting rate decreases. It is interesting that we can -
discovery 0.95
a bump up around 0.85, 6.889which can also1.638 21.530 utilized SVD-combined
be observed in [56].They 2.417
0.90
tensor decomposition to remedy 7.144
incomplete traffic 5.400
data. When filling 22.698 7.973 data
up the fiber-like missing
0.85 7.145 5.415 22.693
(i.e., structural missing data), the RMSE and the missing rate is not a simple linear relationship 7.949 and
a valley is 0.80
presented around 40%. 7.025
Specifically, the3.644
RMSE decreases with 21.425
the increase of 1.917
missing rate
at the first.0.75
However, when missing 6.926 rate is beyond2.184 21.409 with the rise
40%, the RMSE increases 1.841
of missing
0.70 6.955 2.611 21.802
rate. The same trend can be witnessed when extracting rate is lower than 0.85 in our paper. One of 3.710
the reasons 0.65 7.232
is that recovering structural missing data6.698is much more23.763 13.039
difficult than those of random
0.60 7.468 10.180 24.078
missing [57]. It is also probably because of the distribution of critical road sections. 14.537

Figure
Figure 10.
10. Trend
Trend of
of prediction
prediction accuracy
accuracy under different extracting rate.
Sensors 2018, 18, 2287 20 of 26

5.4. Performance for Stochastic and Extreme Cases

In the practical situation, corrupted or missing data generally occur because of monitoring
equipment failure, extreme weather, data transmission errors and so forth and such data weaken the
effectiveness of the prediction model or even disable the model [58]. As stated in [59], data input
failure can be classified into three categories: (a) incidental (random) failures, mainly due to temporary
detection or communication failures; (b) structural failures, caused by long-term equipment failures;
and (c) intrinsic failures, derived from noise, bias and other inherent deviations of detection devices.
For the FCD used in the research, these three types of input failures may occur simultaneously.
In a network, a low traffic volume or signal obstruction caused by very large buildings may result in
unreliable data in certain roads, structural failures will be the primary concern in this paper. Different
from traditional methods that restore missing data by imputation methods before prediction [60],
we predict the traffic conditions of the roads that suffer from structural missing data based on their
adjacent critical road sections.
For the purpose of testing the performance of our model under random structural missing data,
we stochastically extract a part of the road sections at the rate of α, where the value remains the same
as previously mentioned. First, we assign each road section a random value that meets the uniform
distribution with minimum 0 and maximum 1. Then, we sort the random values in an ascending
order and extract the top road sections at the rate of α. In order to eliminate the impact of randomness,
we performed 10 groups of experiments in stochastic case under each extracting rate and the final
result was the average over 10 experiments.
An extreme case is also involved so that the most critical road sections with greater values of
Ci are removed. Table 5 demonstrates the performance of stochastic and extreme case. The drop rate
measures the gap between contrastive scenarios and the case of critical road sections. That is, for drop
rate in Table 6, Index1 of Equation (28) denotes the error index (i.e., RMSE) at the case of critical
road sections and the Index2 indicates the RMSE in the case of stochastic or extreme case at the same
extracting rate of α (α < 1). It can be found that all the prediction accuracies under random missing
data are inferior to the prediction performance of critical road sections the same as in the extreme
case, while the performance is much worse due to the absence of the most valuable spatiotemporal
characteristics. The reason that the margin of values is different for stochastic and extreme case is that
the stochastic case remains more partial spatiotemporal characteristics while the extreme lost the most.
Figure 11 depicts the gap among critical road sections, stochastic case and extreme case. It is obvious
that predicting with the data of critical road sections performs best in all extracting rate, followed by
stochastic case and extreme case.

Table 6. Prediction results in the case of stochastic and extreme situation.

Prediction Performance on Test Dataset

ff Stochastic Case Extreme Case
RMSE Drop Rate (%) RMSE Drop Rate (%)
0.95 7.345 6.619 8.565 24.329
0.90 7.992 11.870 8.679 21.487
0.85 7.837 9.685 8.682 21.512
0.80 7.534 7.245 8.647 23.089
0.75 7.728 11.579 8.734 26.105
0.70 7.412 6.571 8.470 21.783
0.65 7.565 4.605 8.701 20.313
0.60 7.847 5.075 8.785 17.635
 Sensors 2018, 18, 2287 21 of 26
Sensors 2018, 18, x FOR PEER REVIEW 20 of 24

Figure
Figure11.
11. Comparison
Comparison between critical road
between critical roadsections
sections(CRS)
(CRS)case,
case,stochastic
stochastic case
case (SC)
(SC) andand extreme
extreme
case
case(EC).
(EC).

6. Summary and Conclusions

However, although the proposed approach seems to be ineffective under stochastic and extreme
cases, the problems
Structural missingcauseddataby usually
structuralhavemissing data are negative
a massive indeed alleviated,
effect onespecially
short-termin the case of
traffic state
critical road sections, thus validating the effectiveness of our CRS-ConvLSTM.
predictions. In this paper, we propose a ConvLSTM traffic speed prediction method to alleviate the Moreover, the results
present acaused
problem meaningful and worthy
by structural instruction
missing data.forFirst,
trafficthe
management
spatiotemporalthat more monitoring
features devices to
are extracted
should be allocated along critical road sections to improve the level of
measure the impact of road sections on adjacent roads. Spatial weights matrices are constructed toprediction.
represent the topology of the network and the correlation distance is adopted to calculate the degree
6. Summary and Conclusions
of impact between a road section and its local network. Then, correlation distances are integrated by
TOPSIS Structural
to rank missing
the criticaldatadegree
usually of have a massive
all roads. Thus,negative effectsections
critical road on short-term
can be traffic state
distinguished
predictions. In this paper, we propose a ConvLSTM traffic speed prediction
according to their order and subsequently extracted as the input of the prediction model at different method to alleviate
the problem
extracting rates.caused
In theby structuralsection,
prediction missingthe data. First, the
proposed spatiotemporal
ConvLSTM features are
NN is employed toextracted to
forecast future
measure the impact of road sections on adjacent roads. Spatial weights matrices are constructed to
traffic states of the overall network based on the speeds of critical road sections.
represent the topology of the network and the correlation distance is adopted to calculate the degree
To demonstrate the validation of the proposed model, a group of empirical experiments are
of impact between a road section and its local network. Then, correlation distances are integrated
performed using the traffic speed collected at 2 min intervals from a Beijing transportation network
by TOPSIS to rank the critical degree of all roads. Thus, critical road sections can be distinguished
with 278 links. A sensitivity analysis about the different spatial weight matrices is conducted to prove
according to their order and subsequently extracted as the input of the prediction model at different
that the availability of 5-order spatial weight matrix. Compared with the ANN, CNN, LSTM and
extracting rates. In the prediction section, the proposed ConvLSTM NN is employed to forecast future
SAE, our CRS-ConvLSTM achieves an acceptable accuracy with RMSE is of 6.955 at the extracting
traffic states of the overall network based on the speeds of critical road sections.
rate ofTo 0.7. Furthermore,
demonstrate thethe prediction
validation accuracy
of the proposedunder different
model, extracting
a group rates isexperiments
of empirical tested. The results
are
show that in general, the RMSE increases as the extracting rate decreases
performed using the traffic speed collected at 2 min intervals from a Beijing transportation and when the network
top 95% of
road
withsections
278 links. areAused as input,
sensitivity we can
analysis acquire
about the optimal
the different result.
spatial weightA bump
matricesup is
around 0.85 to
conducted is probably
prove
because of the difficulty of recovering structural missing data and the
that the availability of 5-order spatial weight matrix. Compared with the ANN, CNN, LSTM and SAE, distribution of the critical road
sections. However, when
our CRS-ConvLSTM the 𝛼anvalue
achieves is lower
acceptable than 0.65,
accuracy withtheRMSEefficiency of the
is of 6.955 model
at the gradually
extracting rate becomes
of 0.7.
unstable
Furthermore, the prediction accuracy under different extracting rates is tested. The results show thatpart
and unreliable. Consequently, the prediction accuracy is slightly affected when we omit
ofinthe noncritical
general, the RMSEroadincreases
sections as andthean excessiverate
extracting reduction
decreases ofand
inputwhendata istop
the not95%
appropriate either.
of road sections
are Subsequently,
used as input, we a situation
can acquire of stochastic
the optimaland extreme
result. A bump structural
up around missing data
0.85 is in a real
probably scenario
because of is
the difficulty
simulated of recovering
by extracting structural
certain roads missing data andfrom
at rates ranging the distribution
0.60 to 0.95.ofAllthethe
critical road
results in sections.
the context
ofHowever,
randomwhen and the α valuecases
extreme is lower
are than 0.65, the
inferior efficiency
to those of the model
of critical roadgradually
sections.becomes
Although unstable
obvious
and unreliable. Consequently, the prediction accuracy is slightly affected
regularities are not observed for the RMSEs of the model under different extracting rates, the critical when we omit part of the
noncritical
road sections road sections andhave
undoubtedly an excessive reduction of
more prominent input datapower.
predictive is not appropriate
Moreover, either.
the validity of the
spatiotemporal correlation algorithm has also been confirmed.
In summary, we tested the performance of the proposed in a large-scale network, which proved
the availability of the proposed method. In practical implementation, considering that we adopted
deep learning approaches, at least 3 months historical traffic data is necessary to train and test the
prediction model. Before prediction, the spatiotemporal features should be extracted to draw the
critical road sections out of the overall network. Once the structure of the model is determined and
the parameters is trained by historical data, short-term prediction can be conducted based on the
Sensors 2018, 18, 2287 22 of 26

Subsequently, a situation of stochastic and extreme structural missing data in a real scenario is
simulated by extracting certain roads at rates ranging from 0.60 to 0.95. All the results in the context of
random and extreme cases are inferior to those of critical road sections. Although obvious regularities
are not observed for the RMSEs of the model under different extracting rates, the critical road sections
undoubtedly have more prominent predictive power. Moreover, the validity of the spatiotemporal
correlation algorithm has also been confirmed.
In summary, we tested the performance of the proposed in a large-scale network, which proved
the availability of the proposed method. In practical implementation, considering that we adopted
deep learning approaches, at least 3 months historical traffic data is necessary to train and test the
prediction model. Before prediction, the spatiotemporal features should be extracted to draw the
critical road sections out of the overall network. Once the structure of the model is determined and
the parameters is trained by historical data, short-term prediction can be conducted based on the
current states of the critical road sections. Thus, online prediction of the overall traffic network can be
performed by states of critical road sections and the trained model.
For future studies and applications, the quality of data should be addressed since random missing
values affect the prediction accuracy. The imputation method such tensor decomposition is emerging
as a potential approach to remedy missing traffic data. Also, the hierarchy, length and other inherent
attributes of the road should be taken into account when extracting spatial correlations between roads.
Additionally, since weather, temperature, air quality and other external factors affect travel demand,
these factors can be comprehensively considered to improve prediction accuracy. Another strategy for
improving prediction accuracy is the inclusion of more detection devices at critical road sections to
achieve a more reliable prediction.

Author Contributions: H.Y. and Y.R. contributed analysis tools and the idea; G.Y. performed the experiments and
wrote the paper; J.X. was in charge of the final version of the paper; Y.W. collected and processed the data.
Funding: This paper is supported by the National Natural Science Foundation of China (51308021) and Young
Elite Scientist Sponsorship Program by the China Association for Science and Technology.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A

Table A1. Hyperparameters of the ANN when α = 0.70.

Layer Name Parameters Dimensions

0 Input – (195 × 15, )
NN 128 –
1 Activation(relu) – –
Dropout (0.2) – –
NN 256 –
2 Activation(relu) – –
Dropout (0.2) – –
NN 512 –
3 Activation(relu) – –
Dropout (0.2) – –
4 Fully connected 278 (278, )
Sensors 2018, 18, 2287 23 of 26

Table A2. Hyperparameters of the CNN when α = 0.70

Layer Name Parameters Dimensions

0 Input – (1, 195, 15)
Convolution (16, 3, 3) (16, 195, 15)
Pooling (2, 2) (16, 98, 8)
1
Activation(relu) – –
Batch-normalization – –
Convolution (32, 3, 3) (32, 98, 8)
Pooling (2, 2) (32, 49, 4)
2
Activation(relu) – –
Batch-normalization – –
Convolution (64, 3, 3) (64, 49, 4)
Pooling (2, 2) (64, 25, 2)
3
Activation(relu) – –
Batch-normalization – –
Convolution (128, 3, 3) (128, 25, 2)
Pooling (2, 2) (128, 13, 1)
4
Activation(relu) – –
Batch-normalization – –
Convolution (256, 3, 3) (256, 13, 1)
Pooling (2, 2) (256, 7, 1)
5
Activation(relu) – –
Batch-normalization – –
6 Flatten – (1792, )
7 Fully connected 278 (278, )

Table A3. Hyperparameters of the LSTM when α = 0.70

Layer Name Parameters Dimensions

0 Input – (1, 195 × 15)
LSTM 200 –
1
Activation(tanh) – –
LSTM 200 –
2
Activation(tanh) – –
3 Fully connected 278 (278, )

Table A4. Hyperparameters of the SAE when α = 0.70

Task Hidden Layers Hidden Units (Bottom-Top)

30-min traffic prediction 3 [200, 200, 200]

References
1. Elhenawy, M.; Rakha, H. Spatiotemporal traffic state prediction based on discriminatively pre-trained deep
neural networks. Adv. Sci. Technol. Eng. Syst. J. 2017, 2, 678–686. [CrossRef]
2. Huang, W.; Song, G.; Hong, H.; Xie, K. Deep architecture for traffic flow prediction: Deep belief networks
with multitask learning. IEEE Trans. Intell. Transp. Syst. 2014, 15, 2191–2201. [CrossRef]
3. Vlahogianni, E.I.; Karlaftis, M.G.; Golias, J.C. Short-term traffic forecasting: Where we are and where we’re
going. Transp. Res. Part C Emerg. Technol. 2014, 43, 3–19. [CrossRef]
4. Rahmani, M.; Koutsopoulos, H.N.; Ranganathan, A. Requirements and Potential of GPS-Based Floating Car
Data for Traffic Management: Stockholm Case Study. In Proceedings of the International IEEE Conference
on Intelligent Transportation Systems, Funchal, Portugal, 19–22 September 2010; pp. 730–735.
Sensors 2018, 18, 2287 24 of 26

5. Angrisano, A.; Gaglione, S.; Gioia, C. Performance assessment of GPS/GLONASS single point positioning
in an urban environment. Acta Geod. Geophys. 2013, 48, 149–161. [CrossRef]
6. Richards, P.I. Shock waves on highway. Oper. Res. 1956, 4, 42–51. [CrossRef]
7. Chandra, S.R.; Al-Deek, H. Cross-correlation analysis and multivariate prediction of spatial time series of
freeway traffic speeds. Transp. Res. Record J. Transp. Res. Board 2008, 2061, 64–76. [CrossRef]
8. Min, W.; Wynter, L. Real-time road traffic prediction with spatio-temporal correlations. Transp. Res. Part C
Emerg. Technol. 2011, 19, 606–616. [CrossRef]
9. Van Lint, H.; Van Hinsbergen, C. Short-term traffic and travel time prediction models. Transp. Res. E Circ.
2012, E-C168, 22–41.
10. Ahmed, M.S. Analysis of freeway traffic time series data and their application to incident detection.
Equine Vet. Educ. 1979, 6, 32–35.
11. Williams, B.M.; Hoel, L.A. Modeling and forecasting vehicular traffic flow as a seasonal ARIMA process:
Theoretical basis and empirical results. J. Transp. Eng. 2003, 129, 664–672. [CrossRef]
12. Pfeifer, P.; Deutsch, S.J. Identification and interpretation of first order space-time ARMA models. Technometrics
1980, 22, 397–408. [CrossRef]
13. Guo, J.; Huang, W.; Williams, B.M. Adaptive Kalman filter approach for stochastic short-term traffic flow
rate prediction and uncertainty quantification. Transp. Res. Part C 2014, 43, 50–64. [CrossRef]
14. Lint, J.W.C.V. Online learning solutions for freeway travel time prediction. IEEE Trans. Intell. Transp. Syst.
2008, 9, 38–47. [CrossRef]
15. Wang, Y.; Papageorgiou, M.; Messmer, A. Real-time freeway traffic state estimation based on extended
Kalman filter: A case study. Transp. Sci. 2007, 41, 167–181. [CrossRef]
16. Liu, X.; Fang, X.; Qin, Z.; Ye, C.; Xie, M. A short-term forecasting algorithm for network traffic based on
chaos theory and SVM. J. Netw. Syst. Manag. 2011, 19, 427–447. [CrossRef]
17. Sun, Y.; Leng, B.; Guan, W. A novel wavelet-SVM short-time passenger flow prediction in Beijing subway
system. Neurocomputing 2015, 166, 109–121. [CrossRef]
18. Wu, C.H.; Ho, J.M.; Lee, D.T. Travel-time prediction with support vector regression. IEEE Trans. Intell.
Transp. Syst. 2004, 5, 276–281. [CrossRef]
19. Karlaftis, M.G.; Vlahogianni, E.I. Statistical methods versus neural networks in transportation research:
Differences, similarities and some insights. Transp. Res. Part C 2011, 19, 387–399. [CrossRef]
20. Smith, B.L. Forecasting Freeway Traffic Flow for Intelligent Transportation Systems Application; University of
Virginia: Charlottesville, VA, USA, 1995.
21. Smith, B.L.; Demetsky, M.J. Short-term traffic flow prediction: Neural network approach. Transp. Res. Record
1994, 1453, 98–104.
22. Wang, J.; Deng, W.; Guo, Y. New Bayesian combination method for short-term traffic flow forecasting.
Transp. Res. Part C 2014, 43, 79–94. [CrossRef]
23. Jiang, X.; Zhang, L.; Chen, X. Short-term forecasting of high-speed rail demand: A hybrid approach
combining ensemble empirical mode decomposition and gray support vector machine with real-world
applications in China. Transp. Res. Part C Emerg. Technol. 2014, 44, 110–127. [CrossRef]
24. Sun, S.; Zhang, C.; Yu, G. A Bayesian network approach to traffic flow forecasting. IEEE Trans. Intell.
Transp. Syst. 2006, 7, 124–132. [CrossRef]
25. Chen, H.; Grant-Muller, S. Use of sequential learning for short-term traffic flow forecasting. Transp. Res. Part
C Emerg. Technol. 2001, 9, 319–336. [CrossRef]
26. Huang, S.H.; Ran, B. An Application of Neural Network on Traffic Speed Prediction under Adverse Weather
Condition. In Proceedings of the Transportation Research Board Annual Meeting, Washington, DC, USA,
12–16 January 2003.
27. Chan, K.Y.; Dillon, T.S.; Singh, J.; Chang, E. Neural-network-based models for short-term traffic flow
forecasting using a hybrid exponential smoothing and Levenberg–Marquardt algorithm. IEEE Trans. Intell.
Transp. Syst. 2012, 13, 644–654. [CrossRef]
28. Van Hinsbergen, C.I.; Van Lint, J.; Van Zuylen, H. Bayesian committee of neural networks to predict travel
times with confidence intervals. Transp. Res. Part C Emerg. Technol. 2009, 17, 498–509. [CrossRef]
29. Ma, X.; Yu, H.; Wang, Y.; Wang, Y. Large-scale transportation network congestion evolution prediction using
deep learning theory. PLoS ONE 2015, 10, e0119044. [CrossRef] [PubMed]
30. Lecun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436. [CrossRef] [PubMed]
Sensors 2018, 18, 2287 25 of 26

31. Lv, Y.; Duan, Y.; Kang, W.; Li, Z.; Wang, F.Y. Traffic flow prediction with big data: A deep learning approach.
IEEE Trans. Intell. Transp. Syst. 2015, 16, 865–873. [CrossRef]
32. Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [CrossRef]
[PubMed]
33. Ma, X.; Tao, Z.; Wang, Y.; Yu, H.; Wang, Y. Long short-term memory neural network for traffic speed
prediction using remote microwave sensor data. Transp. Res. Part C 2015, 54, 187–197. [CrossRef]
34. Schmidhuber, J. Deep Learning in Neural Networks; Elsevier Science Ltd.: New York, NY, USA, 2015.
35. Zhao, Z.; Chen, W.; Wu, X.; Chen, P.C.Y.; Liu, J. LSTM network: A deep learning approach for short-term
traffic forecast. IET Intell. Transp. Syst. 2017, 11, 68–75. [CrossRef]
36. Ma, X.; Dai, Z.; He, Z.; Ma, J.; Wang, Y.; Wang, Y. Learning traffic as images: A deep convolutional neural
network for large-scale transportation network speed prediction. Sensors 2017, 17, 818. [CrossRef] [PubMed]
37. Cai, P.; Wang, Y.; Lu, G.; Chen, P.; Ding, C.; Sun, J. A spatiotemporal correlative k-nearest neighbor model for
short-term traffic multistep forecasting. Transp. Res. Part C 2016, 62, 21–34. [CrossRef]
38. Habtemichael, F.G.; Cetin, M. Short-term traffic flow rate forecasting based on identifying similar traffic
patterns. Transp. Res. Part C 2016, 66, 61–78. [CrossRef]
39. Wu, Y.; Tan, H. Short-term traffic flow forecasting with spatial-temporal correlation in a hybrid deep learning
framework. arXiv, 2016.
40. Wang, J.; Gu, Q.; Wu, J.; Liu, G.; Xiong, Z. Traffic Speed Prediction and Congestion Source Exploration:
A Deep Learning Method. In Proceedings of the IEEE International Conference on Data Mining, Barcelona,
Spain, 12–15 December 2016; pp. 499–508.
41. Yu, H.; Wu, Z.; Wang, S.; Wang, Y.; Ma, X. Spatiotemporal recurrent convolutional networks for traffic
prediction in transportation networks. Sensors 2017, 17, 1501. [CrossRef] [PubMed]
42. Tan, H.; Feng, G.; Feng, J.; Wang, W.; Zhang, Y.J.; Li, F. A tensor-based method for missing traffic data
completion. Transp. Res. Part C 2013, 28, 15–27. [CrossRef]
43. Chung, F.; Lu, L.; Chung, F.; Chung, F.R.; Graham, F.C.; Chung, K.F. Complex Graphs and Networks; American
Mathematical Society: Providence, RI, USA, 2006; pp. 118–157.
44. Miller, H.J. Tobler’s first law and spatial analysis. Ann. Assoc. Am. Geogr. 2004, 94, 284–289. [CrossRef]
45. Kamarianakis, Y.; Prastacos, P. Space-time modeling of traffic flow. Comput. Geosci. 2005, 31, 119–133.
[CrossRef]
46. Wang, X.; Mueen, A.; Ding, H.; Trajcevski, G.; Scheuermann, P.; Keogh, E. Experimental comparison of
representation methods and distance measures for time series data. Data Min. Knowl. Discov. 2013, 26, 275–309.
[CrossRef]
47. Huang, I.B.; Keisler, J.; Linkov, I. Multi-criteria decision analysis in environmental sciences: Ten years of
applications and trends. Sci. Total Environ. 2011, 409, 3578–3594. [CrossRef] [PubMed]
48. Gong, X.; Lu, Y. Data Mining Based Research on Urban Tide Traffic Problem. In Proceedings of the
International IEEE Conference on Intelligent Transportation Systems, Beijing, China, 12–15 October 2008;
pp. 122–127.
49. Nair, A.S.; Liu, J.C.; Rilett, L.; Gupta, S. Non-Linear Analysis of Traffic Flow. In Proceedings of the 2001 IEEE
Intelligent Transportation Systems, Oakland, CA, USA, 25–29 August 2001; pp. 681–685.
50. Wen, H.; Sun, J.; Zhang, X. Study on traffic congestion patterns of large city in China taking Beijing as
an example. Procedia Soc. Behav. Sci. 2014, 138, 482–491. [CrossRef]
51. Krizhevsky, A.; Sutskever, I.; Hinton, G.E. Imagenet Classification with Deep Convolutional Neural
Networks. In Proceedings of the International Conference on Neural Information Processing Systems,
Lake Tahoe, NV, USA, 3–6 December 2012; pp. 1097–1105.
52. LeCun, Y. Lenet-5, Convolutional Neural Networks. Available online: http://yann.lecun.com/exdb/lenet
(accessed on 12 July 2018).
53. Zeiler, M.D. ADADELTA: An adaptive learning rate method. arXiv 2012, arXiv:1212.5701.
54. Ioffe, S.; Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate
shift. arXiv 2015, arXiv:1502.03167.
55. Caruana, R.; Lawrence, S.; Giles, L. Overfitting in Neural Nets: Backpropagation, Conjugate Gradient,
and Early Stopping. In Proceedings of the International Conference on Neural Information Processing
Systems, Denver, CO, USA, 27–30 November 2000; pp. 381–387.
Sensors 2018, 18, 2287 26 of 26

56. Chen, X.; He, Z.; Wang, J. Spatial-temporal traffic speed patterns discovery and incomplete data recovery via
SVD-combined tensor decomposition. Transp. Res. Part C Emerg. Technol. 2018, 86, 59–77. [CrossRef]
57. Acar, E.; Dunlavy, D.M.; Kolda, T.G.; Mørup, M. Scalable tensor factorizations for incomplete data. Chemom.
Intell. Lab. Syst. 2010, 106, 41–56. [CrossRef]
58. Li, Y.; Li, Z.; Li, L. Missing traffic data: Comparison of imputation methods. Intell. Transp. Syst. IET 2014, 8, 51–57.
[CrossRef]
59. Lint, J.W.C.V.; Hoogendoorn, S.P.; Zuylen, H.J.V. Accurate freeway travel time prediction with state-space
neural networks under missing data. Transp. Res. Part C 2005, 13, 347–369. [CrossRef]
60. Chen, H.; Grant-Muller, S.; Mussone, L.; Montgomery, F. A study of hybrid neural network approaches and
the effects of missing data on traffic forecasting. Neural Comput. Appl. 2001, 10, 277–286. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).