FORECASTING

FORECASTING

• operations planning begins with a forecast of the future demand for

a business organization’s products, which could be goods or services
• what works best in one firm under one set of conditions may be a
complete disaster in another
• generally inaccurate
Forecasting Time Horizons
a forecast is usually classified by the future time horizon
that it covers

3 categories:
1. short-range forecast- less than 3 mos & extends up to 1 yr
-used for planning purchasing, job scheduling, workforce levels,
job assignments & production levels
2. medium-range forecast – up to 3 years
- useful in sales planning, production planning & budgeting, cash
budgeting & analyzing various operating plans
3.long-range forecast – more than 3 years
- used in planning for new products, capital expenditures, facility
location or expansion, and R & D
Types of Forecasts

3 major types of forecasts in planning future operations:

1. Economic forecasts- address the business cycle by

predicting inflation rates, money supplies, housing starts
& other planning indicators
2. Technological forecasts – concerned with rates of
technological progress which can result to the birth of
exciting new products, requiring new plants & equipment
3. Demand forecasts- also called sales forecasts
- drive a company’s production, capacity & scheduling
systems & serve as inputs to financial, marketing &
personnel planning
Qualitative & Quantitative Forecasts
A. Qualitative Forecasting Techniques
- subjective technique
- incorporates the decision maker’s intuition, emotions, personal
experiences, & value system

B. Quantitative Forecasting Techniques

- use a variety of mathematical models
- generally assume that the same underlying causal system that existed in the
past will continue to exist in the future (use of historical data)
Qualitative Forecasts
1. Jury of executive opinion- takes the opinion of a small group of high-level
managers/consultants/experts & results in a group estimate of demand
2. Delphi method- using a group process that allows experts to make forecasts
3. Sales force composite- Based upon sales person’s estimates of expected
sales
4. Consumer market survey- Solicits inputs from customers or potential
customers regarding future purchasing plans
- can suffer from overly optimistic forecasts
Quantitative Forecasts
- use a variety of mathematical models
- generally assume that the same underlying causal system that existed in
the past will continue to exist in the future (use of historical data)
2 models
1. time-series models - naïve approach, moving average, exponential
smoothing, trend projection
2. associative model- linear regression & correlation analysis
Time-series Models

- based on a sequence of evenly spaced data

points
- forecasting time series data implies that
future values are predicted only from
past values & that other variables, no
matter how potentially valuable, may be
ignored
Time-series Models

4 components of time series

1. Trend
2. Seasonality
3. Cycles
4. Random variations
Time-series Models

1. Naïve Approach
- simplest way to forecast
- assume that demand in the next period
will be equal to demand in the most
recent period
Time-series Models

2. Moving Averages
- uses an average of the n most recent periods
of data to forecast the next period

Simple Moving Average

MA = Dn Dn = demand in previous n periods
n n = # of periods in moving average

Weighted Moving Average

WMA =  WnDn Wn = weight for period n
W
Time-series Models

3. Exponential Smoothing
- weighted moving average forecasting technique in which data points are
weighted by an exponential function
- widely used in business & is an important part of many computerized
inventory control systems

Ft = Ft-1 +  (At-1 - Ft-1) Ft = new forecast

0<<1 Ft-1 = previous forecast
 = smoothing or
weighing constant
At-1 = previous periods
actual demand
Time-series Models

Exponential Smoothing w/ Trend Adjustment

FITt = Ft + Tt

Ft = exponentially smoothed forecast

Ft =  (A t-1) + (1 - ) (Ft-1 + Tt-1)

Tt = exponentially smoothed trend

Tt =  (Ft - Ft-1) + (1 - ) Tt-1

 = smoothing constant for the average (0 <  < 1)

 = smoothing constant for the trend (0 <  < 1)
Time-series Models

4. Trend Projection
- fits a trend line to a series of historical
data points & then projects the line into
the future for medium-to-long range
forecasts
- mathematical trend equations: linear,
quadratic, exponential
Time-series Models

Linear Trend Line

- apply the least squares method
- always plot the data to assure of linear
relationship
- can’t predict time periods far beyond
the given database (usually up to 3
periods only)
- deviations around the least squares line
are assumed to be random
Time-series Models

Least Squares Line

ŷ = a + bx
ŷ = computed value of the variable to be predicted (dependent var.)
a = y-axis intercept
b = slope of the regression line (rate of change in y for given changes in x)
x = time (independent var.)
Time-series Models

Slope b = xy – nxaya

x2 – nxa2
x = known values of the independent variable
y = known values of the dependent variable
xa = average of x-values
ya = average of y-values
n = number of data points & or observations
y-intercept a = ya - bxa
Time-series Models

y y-values
dependent actual observations
variable
Trend Line
ŷ = a + bx

Deviation
(error)

x
time
Associative Models

- unlike time-series forecasting, associative

forecasting models usually consider several
variables that are related to the quantity
being predicted
- Regression Analysis
- Correlation Analysis
Associative Models
Multiple Regression Analysis
- allows to build a model with several
independent variables
- regression lines are not “cause-and-
effect” relationships, describe the
relationships among variables
Model: ŷ = a + b1x1 + b2x2
Associative Models
Key Terms:
1. Intercept
2. R square – measures the proportion of the
variation in the dependent variable that is
explained by the multiple regression equation
3. Coefficients – tells whether independent
variables have positive or negative relationship
with the dependent variable
4. T-stat – value of the coefficient estimate divided
by its standard error
5. P-value – explains the variability of Q
6. F, significance F – the total validity of the model
Associative Models
Standard Error of the Estimate
- measures the accuracy of the regression estimates
- called the standard deviation of the regression
- measures the error from the dependent variable,
y, to the regression line
(y – yc)
Sy,x =
n-2
y = y-value of each data point
yc = computed value of the dependent variable, from
the regression equation
n = number of data points
Associative Models

Correlation Analysis
- another way to evaluate the
relationship between 2 variables
- expresses the degree or strength of the
linear relationship
- -1 < coefficient of correlation < +1

r = nxy - xy______
 [nx2 - (x)2] [ny2 - (y)2]
Associative Models

Four Values of the Correlation Coefficient

1. Perfect positive correlation, r = +1
y

x
Associative Models

2. Positive Correlation, 0 < r < 1

x
Associative Models

3. No correlation, r = 0

x
Associative Models

4. Perfect negative correlation, r = -1

x
Forecast with Seasonal Variations
- seasonal variations in data are regular up &
down
- movements in a time series that relate to
recurring events such as weather & holidays
- seasonality applied to recurring patterns
(hourly,weekly, monthly…)
- important for capacity planning in
organizations that handle peak loads:
banks during Friday afternoons,
transportation companies during rush
hours & certain occasions, fast foods
Forecast with Seasonal Variations

Steps to get seasonal forecast

1. Get average of all data points
2. Divide each data point by the average
3. Get the average for each period
Forecast Evaluation

- some forecasting techniques will provide

more accuracy than others in a given
situation
- forecast error is the difference between the
value that occurs & the value that was
predicted for a given time period

positive errors - forecast errors are too low

negative errors- forecast errors are too high

Error e=F–D
et = Ft - Dt t in certain period
Forecast Evaluation

3 Methods
1. Mean Absolute Deviation (MAD)
MAD = (1/n)   ei  from i= 1 to n

2. Mean Square Error (MSE)

MSE = (1/n)  ei 2 from i= 1 to n

3. Mean Absolute Percentage Error (MAPE)

MAPE = (1/n)   ei /Di x 100 from i= 1 to n