gedovzo2h

March 9, 2025

0.1 1. Importing Required Dependencies

[455]: import requests as re
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.preprocessing import StandardScaler, LabelEncoder
from sklearn.decomposition import PCA
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix, accuracy_score
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier

0.2 2. Importing and Exploring Data

[456]: # data importing
wine = pd.read_csv("C:\\Users\\Gautham\\Downloads\\winequality-red-mod.csv")
wine.head()

[456]: id fixed acidity volatile acidity citric acid residual sugar \

0 1 7.4 0.70 0.00 1.9
1 2 7.8 0.88 0.00 2.6
2 3 7.8 0.76 0.04 2.3
3 4 11.2 0.28 0.56 1.9
4 5 7.4 0.70 0.00 1.9

flavonoids chlorides free sulfur dioxide total sulfur dioxide density \

0 0.590 0.076 11.0 34.0 0.9978
1 0.754 0.098 25.0 67.0 0.9968
2 0.688 0.092 15.0 54.0 0.9970
3 0.459 0.075 17.0 60.0 0.9980
4 0.590 0.076 11.0 34.0 0.9978

pH sulphates alcohol country quality

0 3.51 0.56 9.4 UK 5

1
 1 3.20 0.68 9.8 Italy 5
2 3.26 0.65 9.8 Italy 5
3 3.16 0.58 9.8 Spain 6
4 3.51 0.56 9.4 UK 5

[457]: # data exploring

print(wine.shape)
print(wine.ndim)
wine.info()

(1506, 15)
2
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1506 entries, 0 to 1505
Data columns (total 15 columns):
id 1506 non-null int64
fixed acidity 1504 non-null float64
volatile acidity 1504 non-null float64
citric acid 1504 non-null float64
residual sugar 1504 non-null float64
flavonoids 1504 non-null float64
chlorides 1504 non-null float64
free sulfur dioxide 1504 non-null float64
total sulfur dioxide 1504 non-null float64
density 1504 non-null float64
pH 1504 non-null float64
sulphates 1504 non-null float64
alcohol 1504 non-null float64
country 1504 non-null object
quality 1506 non-null int64
dtypes: float64(12), int64(2), object(1)
memory usage: 170.6+ KB

0.3 3. Cleaning Data

Two rows are missing categorical data and it is better to drop them since we cannot find the mean
values of categorical data.
[458]: # dropping 2 rows with nan since there is no categorical data available
wine = wine.dropna()
wine.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1504 entries, 0 to 1505
Data columns (total 15 columns):
id 1504 non-null int64
fixed acidity 1504 non-null float64
volatile acidity 1504 non-null float64

2
 citric acid 1504 non-null float64
residual sugar 1504 non-null float64
flavonoids 1504 non-null float64
chlorides 1504 non-null float64
free sulfur dioxide 1504 non-null float64
total sulfur dioxide 1504 non-null float64
density 1504 non-null float64
pH 1504 non-null float64
sulphates 1504 non-null float64
alcohol 1504 non-null float64
country 1504 non-null object
quality 1504 non-null int64
dtypes: float64(12), int64(2), object(1)
memory usage: 182.1+ KB

[459]: wine.describe()

[459]: id fixed acidity volatile acidity citric acid \

count 1504.000000 1504.000000 1504.000000 1504.000000
mean 752.880984 8.406981 0.524890 0.273910
std 434.843358 1.757966 0.183892 0.198186
min 1.000000 -1.000000 -1.000000 -1.000000
25% 376.750000 7.200000 0.390000 0.100000
50% 752.500000 8.000000 0.520000 0.260000
75% 1128.250000 9.300000 0.635000 0.430000
max 1506.000000 15.900000 1.580000 1.000000

residual sugar flavonoids chlorides free sulfur dioxide \

count 1504.000000 1504.000000 1504.000000 1504.000000
mean 2.540060 0.597992 0.087533 15.597074
std 1.406215 0.252626 0.055560 10.456825
min -1.000000 -1.000000 -1.000000 -1.000000
25% 1.900000 0.494000 0.071000 7.000000
50% 2.200000 0.559000 0.080000 13.000000
75% 2.600000 0.633000 0.091000 21.000000
max 15.500000 3.206000 0.611000 72.000000

total sulfur dioxide density pH sulphates \

count 1504.000000 1504.000000 1504.000000 1504.000000
mean 46.764628 0.996155 3.302553 0.657606
std 33.240872 0.057661 0.190515 0.177967
min -1.000000 -1.000000 -1.000000 -1.000000
25% 22.000000 0.995700 3.200000 0.550000
50% 38.000000 0.996800 3.300000 0.620000
75% 63.000000 0.997900 3.400000 0.730000
max 289.000000 1.999400 4.010000 2.000000

3
 alcohol quality
count 1504.000000 1504.000000
mean 10.427238 5.635638
std 1.434245 0.815816
min -1.000000 3.000000
25% 9.500000 5.000000
50% 10.100000 6.000000
75% 11.100000 6.000000
max 45.300000 8.000000

0.4 4. Data Visualisation

[460]: #Correlation Heatmap of Variables
plt.figure(figsize=(10,8))
sns.heatmap(wine.corr(),cmap='YlGnBu',annot=True,cbar=False)

[460]: <matplotlib.axes._subplots.AxesSubplot at 0x68febef0>

4
[461]: # plotting the quality of wine using a countplot
fig = plt.figure(figsize = (8,4))
sns.set_style("whitegrid")
sns.countplot(wine['quality'],palette='plasma')
plt.title("QUALITY OF WINE", size=18)

[461]: Text(0.5,1,'QUALITY OF WINE')

[462]: #plotting the wine from different countries using a countplot

fig = plt.figure(figsize = (8,4))
sns.set_style("whitegrid")
sns.countplot(wine['country'],palette='plasma')
plt.title("WINE FROM COUNTRIES", size=18)

[462]: Text(0.5,1,'WINE FROM COUNTRIES')

5
[463]: # plotting the quality of wine against citric acid using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'citric acid', data=wine, palette="Spectral")
plt.title('Quality vs Citric Acid', size=20)

[463]: Text(0.5,1,'Quality vs Citric Acid')

6
[464]: # plotting the quality of wine against residual sugar using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'residual sugar', data=wine, palette="Spectral")
plt.title('Quality vs Residual Sugar', size=20)

[464]: Text(0.5,1,'Quality vs Residual Sugar')

7
[465]: # plotting the quality of wine against flavonoids using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'flavonoids', data=wine, palette="Spectral")
plt.title('Quality vs Flavonoids', size=20)

[465]: Text(0.5,1,'Quality vs Flavonoids')

8
[466]: # plotting the quality of wine against chlorides using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'chlorides', data=wine, palette='Spectral')
plt.title('Quality vs Chlorides', size=20)

[466]: Text(0.5,1,'Quality vs Chlorides')

9
[467]: # plotting the quality of wine against free sulphur dioxide using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'free sulfur dioxide', data=wine, palette='Spectral')
plt.title('Quality vs Free Sulphur Dioxide', size=20)

[467]: Text(0.5,1,'Quality vs Free Sulphur Dioxide')

10
[468]: # plotting the quality of wine against total sulphur dioxide using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'total sulfur dioxide', data=wine, palette='Spectral')
plt.title('Quality vs Total Sulphur Dioxide', size=20)

[468]: Text(0.5,1,'Quality vs Total Sulphur Dioxide')

11
[469]: # plotting the quality of wine against the density using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'density', data=wine, palette='Spectral')
plt.title('Quality vs Density', size=20)

[469]: Text(0.5,1,'Quality vs Density')

12
[470]: # plotting the quality of wine against PH level using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'pH', data=wine, palette='Spectral')
plt.title('Quality vs PH Level', size=20)

[470]: Text(0.5,1,'Quality vs PH Level')

13
[471]: # plotting the quality of wine against Sulphates using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'sulphates', data=wine, palette='Spectral')
plt.title('Quality vs Sulphates', size=20)

[471]: Text(0.5,1,'Quality vs Sulphates')

14
[472]: # plotting the quality of wine against the alcohol content using a barplot
sns.set_style("whitegrid")
sns.barplot('quality', 'alcohol', data=wine, palette='Spectral')
plt.title('Quality vs Alcohol', size=20)

[472]: Text(0.5,1,'Quality vs Alcohol')

15
 0.5 5. PreProcessing Data for Building Machine Learning Algorithm
[473]: #Creating binary classificaion for prediction variable.
#classifyin wine as good and bad by giving a limit for the quality
bins = (2, 6.5, 8)
review = ['bad', 'good']
wine['quality'] = pd.cut(wine['quality'], bins = bins, labels = review)

[474]: #assigning labels to quality variable

label_quality = LabelEncoder()

#0 is bad review and 1 is good review

wine['quality'] = label_quality.fit_transform(wine['quality'])

#countplot for quality variable

sns.countplot(wine['quality'], palette='colorblind')

[474]: <matplotlib.axes._subplots.AxesSubplot at 0x61547970>

16
[475]: ## replacing the categorical variable to an integer
country_to_nums = {'country': {'UK': 1, 'Italy': 2, 'Spain':3}}
wine.replace(country_to_nums, inplace=True)
wine.head()

[475]: id fixed acidity volatile acidity citric acid residual sugar \

0 1 7.4 0.70 0.00 1.9
1 2 7.8 0.88 0.00 2.6
2 3 7.8 0.76 0.04 2.3
3 4 11.2 0.28 0.56 1.9
4 5 7.4 0.70 0.00 1.9

flavonoids chlorides free sulfur dioxide total sulfur dioxide density \

0 0.590 0.076 11.0 34.0 0.9978
1 0.754 0.098 25.0 67.0 0.9968
2 0.688 0.092 15.0 54.0 0.9970
3 0.459 0.075 17.0 60.0 0.9980
4 0.590 0.076 11.0 34.0 0.9978

pH sulphates alcohol country quality

0 3.51 0.56 9.4 1 0
1 3.20 0.68 9.8 2 0
2 3.26 0.65 9.8 2 0
3 3.16 0.58 9.8 3 0

17
 4 3.51 0.56 9.4 1 0

[476]: # dividing the dataset as prediction variable and feature variabes

x = wine.drop('quality', axis = 1)
y = wine['quality']

0.6 6. PCA Analysis

[549]: #Scaling the data using StandardScalar
sc = StandardScaler()
x = sc.fit_transform(x)

[550]: # Performing PCA

pca = PCA()
x_pca = pca.fit_transform(x)

#plot the graph to find the principal components

sns.set_style("whitegrid")
plt.figure(figsize=(7,7))
plt.plot(np.cumsum(pca.explained_variance_ratio_), 'ro-')

[550]: [<matplotlib.lines.Line2D at 0x5fc81fb0>]

18
[551]: #from the plot above, we can see that 9 principal components attribute for 90%␣
↪of variation in the data.

#we can pick the first 9 components for our prediction.

pca_n = PCA(n_components=9)
x_pca_new = pca_n.fit_transform(x)

[552]: # Splitting the data into train and test

x_train, x_test, y_train, y_test = train_test_split(x_pca_new, y, test_size = 0.
↪2)

[553]: print(x_train.shape)
print(x_test.shape)
print(y_train.shape)
print(y_test.shape)

19
 (1203, 9)
(301, 9)
(1203,)
(301,)

0.7 7. Logistic Regression

[554]: lr = LogisticRegression()
lr.fit(x_train, y_train)
lr_predict = lr.predict(x_test)

c:\users\gautham\python\python36-32\lib\site-
packages\sklearn\linear_model\logistic.py:433: FutureWarning: Default solver
will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.
FutureWarning)

[555]: lr_conf_matrix = confusion_matrix(y_test, lr_predict)

lr_acc_score = accuracy_score(y_test, lr_predict)
print(lr_conf_matrix)
print(lr_acc_score*100)

[[261 10]
[ 18 12]]
90.69767441860465

0.8 8. Decision Tree Classifier

[556]: dt = DecisionTreeClassifier()
dt.fit(x_train,y_train)
dt_predict = dt.predict(x_test)

[557]: dt_conf_matrix = confusion_matrix(y_test, dt_predict)

dt_acc_score = accuracy_score(y_test, dt_predict)
print(dt_conf_matrix)
print(dt_acc_score*100)

[[245 26]
[ 14 16]]
86.71096345514951

0.9 9. Random Forest Classifier

[558]: rf = RandomForestClassifier()
rf.fit(x_train, y_train)
rf_predict=rf.predict(x_test)

c:\users\gautham\python\python36-32\lib\site-
packages\sklearn\ensemble\forest.py:246: FutureWarning: The default value of

20
 n_estimators will change from 10 in version 0.20 to 100 in 0.22.
"10 in version 0.20 to 100 in 0.22.", FutureWarning)

[559]: rf_conf_matrix = confusion_matrix(y_test, rf_predict)

rf_acc_score = accuracy_score(y_test, rf_predict)
print(rf_conf_matrix)
print(rf_acc_score*100)

[[264 7]
[ 17 13]]
92.02657807308971

0.10 10. Accuracy Comparison

[560]: ## Creating a datafram to compare the accuracy of the three different␣
↪algorithms performed.

acScore = pd.DataFrame()
acScore['Model'] = ['Linear Regression', 'Decision Tree', 'Random Forest␣
↪Classifier']

ac1 = lr_acc_score*100
ac2 = dt_acc_score*100
ac3 = rf_acc_score*100
acScore['Score'] = [ac1,ac2,ac3]
acScore

[560]: Model Score

0 Linear Regression 90.697674
1 Decision Tree 86.710963
2 Random Forest Classifier 92.026578

[561]: # plotting the accuracy of algorithms using a barplot

sns.set_style("whitegrid")
plot = sns.catplot(x='Model', y='Score', kind='bar', data=acScore,␣
↪palette='Spectral')

# plot.ax gives the axis object

# plot.ax.patches gives list of bars that can be access using index starting at␣
↪0

for i, bar in enumerate(plot.ax.patches):

h = bar.get_height()
plot.ax.text(
i, # bar index (x coordinate of text)
h+10, # y coordinate of text
'{}'.format(int(h)), # y label
ha='center',
va='center',

21
 fontweight='bold',
size=20)

0.11 11. Conclusion

Through visualisation, it shows that Italy produces the most number of wines, followed by Spain
and UK.
While predicting the quality of wine, random forest classifier algorithm produced the best results
with 92% accuracy, followed by Logistic Regression with 90% accuracy and Decision Tree Classifier
with 86% accuracy.

22