Source

Wine Quality Data Set: Document Flow

i. Dataset information

1. Data Injestion

2. EDA

3. Preprocessing

4. Model Building (SVC/SVR)

5. Evaluation

Abstract: Two datasets are included, related to red and white vinho verde wine samples, from the north of
Portugal. The goal is to model wine quality based on physicochemical tests.

The two datasets are related to red and white variants of the Portuguese "Vinho Verde" wine. For more details,
consult:. Due to privacy and logistic issues, only physicochemical (inputs) and sensory (the output) variables
are available (e.g. there is no data about grape types, wine brand, wine selling price, etc.).

Scope of Work:

These datasets can be viewed as classification or regression tasks.

1. The classes are ordered and not balanced (e.g. there are many more normal wines than excellent or poor
ones).
2. Outlier detection algorithms could be used to detect the few excellent or poor wines.
3. We are not sure if all input variables are relevant. So it could be interesting to test feature selection
methods.

Attribute Information:

Input variables (based on physicochemical tests):

1. Feature columns

* fixed acidity | Continous Data

* volatile acidity | Continous Data

* citric acid | Continous Data

* residual sugar | Continous Data

* chlorides | Continous Data

* free sulfur dioxide | Continous Data

* total sulfur dioxide | Continous Data

* density | Continous Data

* pH | Continous Data

*  sulphates | Continous Data

*  alcohol | Continous Data

2. Target column

* quality | Ordinal data (score between 3 to 8)

Citation:
P. Cortez, A. Cerdeira, F. Almeida, T. Matos and J. Reis.
Modeling wine preferences by data mining
from physicochemical properties. In Decision Support Systems, Elsevier, 47(4):547-553, 2009.

1. Data Injestion

Library Import

1 import pandas as pd

2 import numpy as np

3 from sklearn.model_selection import train_test_split, GridSearchCV

4 from sklearn.metrics import classification_report, accuracy_score, confusion_matrix , roc_auc_score, roc_curve

5 from sklearn.preprocessing import StandardScaler

 6 import matplotlib.pyplot as plt

7 plt.style.use('ggplot')

8 import seaborn as sns

9 import warnings

10
warnings.filterwarnings("ignore")

11
%matplotlib inline

12

13

14
# from pandas_profiling import ProfileReport

15
 
16
# ! pip install https://github.com/pandas-profiling/pandas-profiling/archive/master.zip 

Data Import

1 data = pd.read_csv('https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-red.csv', header = 0, sep=';')

2 data.head()

fixed volatile citric residual free sulfur total sulfur

chlorides density pH sulphates alcohol qu
acidity acidity acid sugar dioxide dioxide

0 7.4 0.70 0.00 1.9 0.076 11.0 34.0 0.9978 3.51 0.56 9.4

1 7.8 0.88 0.00 2.6 0.098 25.0 67.0 0.9968 3.20 0.68 9.8

2 7.8 0.76 0.04 2.3 0.092 15.0 54.0 0.9970 3.26 0.65 9.8

3 11.2 0.28 0.56 1.9 0.075 17.0 60.0 0.9980 3.16 0.58 9.8

4 7.4 0.70 0.00 1.9 0.076 11.0 34.0 0.9978 3.51 0.56 9.4

EDA

1 data.columns

Index(['fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar',

'chlorides', 'free sulfur dioxide', 'total sulfur dioxide', 'density',

'pH', 'sulphates', 'alcohol', 'quality'],

dtype='object')

1 data.info()

<class 'pandas.core.frame.DataFrame'>

RangeIndex: 1599 entries, 0 to 1598

Data columns (total 12 columns):

# Column Non-Null Count Dtype

--- ------ -------------- -----

0 fixed acidity 1599 non-null float64

1 volatile acidity 1599 non-null float64

2 citric acid 1599 non-null float64

3 residual sugar 1599 non-null float64

4 chlorides 1599 non-null float64

5 free sulfur dioxide 1599 non-null float64

6 total sulfur dioxide 1599 non-null float64

7 density 1599 non-null float64

 8 pH 1599 non-null float64

9 sulphates 1599 non-null float64

10 alcohol 1599 non-null float64

11 quality 1599 non-null int64

dtypes: float64(11), int64(1)

memory usage: 150.0 KB

Shape of dataset

1 data.shape

(1599, 12)

Summary of dataset

1 data.describe().T

count mean std min 25% 50% 75% max

fixed acidity 1599.0 8.319637 1.741096 4.60000 7.1000 7.90000 9.200000 15.90000

volatile acidity 1599.0 0.527821 0.179060 0.12000 0.3900 0.52000 0.640000 1.58000

citric acid 1599.0 0.270976 0.194801 0.00000 0.0900 0.26000 0.420000 1.00000

residual sugar 1599.0 2.538806 1.409928 0.90000 1.9000 2.20000 2.600000 15.50000

chlorides 1599.0 0.087467 0.047065 0.01200 0.0700 0.07900 0.090000 0.61100

free sulfur dioxide 1599.0 15.874922 10.460157 1.00000 7.0000 14.00000 21.000000 72.00000

total sulfur dioxide 1599.0 46.467792 32.895324 6.00000 22.0000 38.00000 62.000000 289.00000

density 1599.0 0.996747 0.001887 0.99007 0.9956 0.99675 0.997835 1.00369

pH 1599.0 3.311113 0.154386 2.74000 3.2100 3.31000 3.400000 4.01000

sulphates 1599.0 0.658149 0.169507 0.33000 0.5500 0.62000 0.730000 2.00000

alcohol 1599.0 10.422983 1.065668 8.40000 9.5000 10.20000 11.100000 14.90000

quality 1599.0 5.636023 0.807569 3.00000 5.0000 6.00000 6.000000 8.00000

Checking for null values

1 data.isna().sum()

fixed acidity 0

volatile acidity 0

citric acid 0

residual sugar 0

chlorides 0

free sulfur dioxide 0

total sulfur dioxide 0

density 0

 pH 0

sulphates 0

alcohol 0

quality 0

dtype: int64

Proportion of count data on "quality" columns

1 data.quality.unique()

2 round(data.quality.value_counts()/(len(data))*100,2)

5 42.59

6 39.90

7 12.45

4 3.31
8 1.13
3 0.63
Name: quality, dtype: float64

Obesrvation:

1. Data set is imbalanced

Univariate analysis

Numerical Columns

1 plt.figure(figsize=(15, 15))

2 plt.suptitle('Univariate Analysis of Numerical Features', fontsize=20, fontweight='bold', alpha=0.8, y=1.)

3 numerical_features = [col for col in data.columns if data[col].dtypes != 'O']

4
5 for i in range(0, len(numerical_features)):

6     plt.subplot(5, 3, i+1)

7     sns.kdeplot(x=data[numerical_features[i]],shade=True, color='b')

8     plt.xlabel(numerical_features[i])

9     plt.tight_layout()

Observations:

1. Normally distributed columns: 'density', 'pH' .

2. Column with Outliers: 'Residual Sugar', 'Chlorides'
3. Left skewed Columns: 'Fixed acidity', 'Free sulphur dioxide', 'Total sulphur dioxide', 'sulphates', 'alcohol'
4. Right skewed Columns: No such column
5. Columns with uneven distribution: Volatile acidic, citric acid

Multivariate analysis

Multivariate analysis is the analysis of more than one variable.

Checking for multicolinearity

1 # In numerical column
2 plt.figure(figsize = (15,10))
3 matrix = np.triu(data.corr())
 4 sns.heatmap(data.corr(), annot=True, mask = matrix)
5 plt.yticks(rotation=45)
6 plt.show()

Observations:

1. Very highly correlated columns (corr > .90): None

2. High correlated columns (corr > 90): None

Relation between feature and label column

1 feature = data.drop(columns = 'quality')

2 feature.columns

Index(['fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar',

'chlorides', 'free sulfur dioxide', 'total sulfur dioxide', 'density',

 'pH', 'sulphates', 'alcohol'],

dtype='object')

Here we will make an approach to understand the relation between the feature and target columns.

Feature columns are:

'fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar', 'chlorides', 'free sulfur dioxide', 'total sulfur
dioxide', 'density','pH', 'sulphates', 'alcohol'

Label columns:

'quality'

1 # Getting the percentage of each category in 'quality' column

2 sns.countplot(x = 'quality', data= data)  

<matplotlib.axes._subplots.AxesSubplot at 0x7f4a042035d0>

1 feature_continous = [col for col in feature.columns if data[col].dtypes != 'O']

1 fig = plt.figure(figsize=(15, 50))
2 plt.suptitle('Box Plot between feature and continous label ', fontsize = 20, y = 1)

3
4 for i in range(0, len(feature_continous)):

5     ax = plt.subplot(10, 3, i+1)

6     sns.boxplot(data = data, x = 'quality', y = data[feature_continous[i]])

7     plt.tight_layout()

Preliminary Conclusions:

1. Increasing sulphates will imporove the quality of wine.

2. Increasing alcohol will imporove the quality of wine.
3. Decreasing volatilic acid will imporove the quality of wine.
4. Increasing citric acid will imporove the quality of wine.

1 sns.pairplot(data)

 <seaborn.axisgrid.PairGrid at 0x7f4a04037610>

Preprocessing
Bringing down the data to same scale will surely reduce the computation time

1 x = data.drop(columns = 'quality')

2 y = data['quality']

1 x_train, x_test, y_train,y_test = train_test_split(x , y, test_size = .2, random_state = 0)

1 scaler = StandardScaler()

2 x_train_tf = scaler.fit_transform(x_train)

3 # get the parameter 

4 scaler.mean_

array([ 8.33737295, 0.53 , 0.27218139, 2.55316654, 0.08702424,

16.03283815, 47.11415168, 0.99675746, 3.30982799, 0.6590774 ,

10.41399531])

1 x_test_tf = scaler.transform(x_test)

Model Building

1 from sklearn.svm import SVC 

Raw SVC

1 model_svc = SVC()

2 model_svc.fit(x_train_tf, y_train)

3 model_svc.score(x_train_tf, y_train)

4 print(f"Accuracy score is: {model_svc.score(x_train_tf, y_train)}") 

5 predict_raw = model_svc.predict(x_test_tf)

6
7
8

Accuracy score is: 0.6669272869429241

Tuning the SVC

1 model_svc_tune = SVC()

2 params = [{'C': [.5,.9,1,1.2,1.3,1.5]}]

3 clf = GridSearchCV(model_svc_tune, params, cv = 10, scoring='accuracy')

4 clf.fit(x_train_tf, y_train)

5 print(f'best value of C is {clf.best_params_}')

7 model_svc_tune = SVC()

8 params = {'kernel': [  'rbf','linear','poly','sigmoid' ],

9            'degree': [  2,3,4,5,6 ]}

10
clf = GridSearchCV(model_svc_tune, params, cv = 10, scoring='accuracy')

11
clf.fit(x_train_tf, y_train)

12
print(clf.best_params_)

13

14
model_svc_tune = SVC()

15
params = {'gamma' :[0.8,0.9,1,1.1,1.2,1.3]}

16
clf = GridSearchCV(model_svc_tune, params, cv = 10, scoring='accuracy')

17
clf.fit(x_train_tf, y_train)

18
print(clf.best_params_)

best value of C is {'C': 1.2}

{'degree': 2, 'kernel': 'rbf'}

{'gamma': 1}

1 params = {

2     'C': [.9,1,1.2,1.3],

3     'kernel':['rbf','linear'],

4     'gamma': [.9,1,1.1]

5 }

6
7 clf = GridSearchCV(model_svc_tune, params, cv = 10, scoring='accuracy')

8 clf.fit(x_train_tf, y_train)

9 print(clf.best_params_)

{'C': 1.3, 'gamma': 1.1, 'kernel': 'rbf'}

1 model_svc_tune = SVC(C = 1.3,kernel= 'rbf', gamma = 1.3)

2 model_svc_tune.fit(x_train_tf, y_train)

3 predict_tuned = model_svc_tune.predict(x_test_tf)

Evaluation

Raw Model

1 print(f'Accuracy Score: {accuracy_score(y_test, predict_raw)}')

Accuracy Score: 0.64375

1 print('Classification report')

2 print(classification_report(y_test, predict_raw))

Classification report

precision recall f1-score support

 3 0.00 0.00 0.00 2

4 0.00 0.00 0.00 11

5 0.67 0.75 0.71 135

6 0.63 0.68 0.66 142
7 0.50 0.30 0.37 27

8 0.00 0.00 0.00 3

accuracy 0.64 320

macro avg 0.30 0.29 0.29 320

weighted avg 0.61 0.64 0.62 320

Tuned Model

1 print(f'Accuracy Score: {accuracy_score(y_test, predict_tuned)}')

Accuracy Score: 0.675

1 print('Classification Report')

2 print(classification_report(y_test, predict_tuned))

Classification Report

precision recall f1-score support

3 0.00 0.00 0.00 2

4 0.00 0.00 0.00 11

5 0.66 0.79 0.72 135

6 0.70 0.68 0.69 142
7 0.71 0.44 0.55 27

8 0.00 0.00 0.00 3

accuracy 0.68 320

macro avg 0.34 0.32 0.33 320

weighted avg 0.65 0.68 0.66 320

1 pd.read_csv("https://raw.githubusercontent.com/srinivasav22/Graduate-Admission-Prediction/master/Admission_Predict_Ver1.1.csv")

Colab paid products

-
Cancel contracts here