8/21/2020 Course 2 - Final Project

About the dataset - Car Evaluation Dataset (UCI Machine Learning

Repository
Car Evaluation Database was derived from a simple hierarchical decision model originally developed for the
demonstration of DEX, M. Bohanec, V. Rajkovic: Expert system for decision making. Sistemica 1(1), pp. 145-
157, 1990.). The model evaluates cars according to the following concept structure:

CAR car acceptability . PRICE overall price . . buying buying price . . maint price of the maintenance . TECH
technical characteristics . . COMFORT comfort . . . doors number of doors . . . persons capacity in terms of
persons to carry . . . lug_boot the size of luggage boot . . safety estimated safety of the car

1. Number of Instances: 1728 (instances completely cover the attribute space)

2. Number of Attributes: 6
3. Attribute Values:

buying -> v-high, high, med, low maint -> v-high, high, med, low doors -> 2, 3, 4, 5-more persons -> 2, 4,
more lug_boot -> small, med, big safety -> low, med, high
4. Missing Attribute Values: none

In [19]: import numpy as np

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

Data has been modified and stored in an excel sheet, and we will convert string values to integer values to
implement scikit packages

In [20]: data = pd.read_excel('car_data1.xlsx')

In [21]: data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1728 entries, 0 to 1727
Data columns (total 8 columns):
Unnamed: 0 1728 non-null int64
buying 1728 non-null object
maint 1728 non-null object
doors 1728 non-null object
persons 1728 non-null object
lug_boot 1728 non-null object
safety 1728 non-null object
class 1728 non-null object
dtypes: int64(1), object(7)
memory usage: 108.1+ KB

localhost:8888/nbconvert/html/Desktop/Python Data Products for Predictive Anlaytics Specialization/Course 2 - Design Thinking and Predictive Analytics for Data P… 1/5
8/21/2020 Course 2 - Final Project

To check for null values^

In [22]: data.head()

Out[22]:
Unnamed: 0 buying maint doors persons lug_boot safety class

0 0 vhigh vhigh 2 2 small low unacc

1 1 vhigh vhigh 2 2 small med unacc

2 2 vhigh vhigh 2 2 small high unacc

3 3 vhigh vhigh 2 2 med low unacc

4 4 vhigh vhigh 2 2 med med unacc

In [23]: # Now checking for unique values

In [24]: for i in data.columns:

print(data[i].unique(),"\t",data[i].nunique())

[ 0 1 2 ... 1725 1726 1727] 1728

['vhigh' 'high' 'med' 'low'] 4
['vhigh' 'high' 'med' 'low'] 4
['2' '3' '4' '5more'] 4
['2' '4' 'more'] 3
['small' 'med' 'big'] 3
['low' 'med' 'high'] 3
['unacc' 'acc' 'vgood' 'good'] 4

In [25]: from sklearn.preprocessing import LabelEncoder

# scikit packages work well with integers, and I faced a few issues as t
he current dataset has string values
# converted string values to integer

In [26]: le=LabelEncoder()
for i in data.columns:
data[i]=le.fit_transform(data[i])

In [27]: data.head()

Out[27]:
Unnamed: 0 buying maint doors persons lug_boot safety class

0 0 3 3 0 0 2 1 2

1 1 3 3 0 0 2 2 2

2 2 3 3 0 0 2 0 2

3 3 3 3 0 0 1 1 2

4 4 3 3 0 0 1 2 2

localhost:8888/nbconvert/html/Desktop/Python Data Products for Predictive Anlaytics Specialization/Course 2 - Design Thinking and Predictive Analytics for Data P… 2/5
8/21/2020 Course 2 - Final Project

We also create a heatmap of columns of dataset. It shows us Pearson's correlation coefficient as well.

In [29]: fig=plt.figure(figsize=(10,6))
sns.heatmap(data.corr(),annot=True).set_title("Heatmap showing Pearson's
correlation coefficient")

Out[29]: Text(0.5, 1, "Heatmap showing Pearson's correlation coefficient")

X dataframe consists of input data and features and y dataframe is the series which has results that we will try
to predict

In [30]: X = data[data.columns[:-1]]
y = data['class']

In [31]: data.head()

Out[31]:
Unnamed: 0 buying maint doors persons lug_boot safety class

0 0 3 3 0 0 2 1 2

1 1 3 3 0 0 2 2 2

2 2 3 3 0 0 2 0 2

3 3 3 3 0 0 1 1 2

4 4 3 3 0 0 1 2 2

In [32]: # We divide our data into train and test splits

from sklearn.model_selection import train_test_split

localhost:8888/nbconvert/html/Desktop/Python Data Products for Predictive Anlaytics Specialization/Course 2 - Design Thinking and Predictive Analytics for Data P… 3/5
8/21/2020 Course 2 - Final Project

In [33]: X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3,

random_state=42)

In [35]: from sklearn.linear_model import LogisticRegression

from sklearn.model_selection import train_test_split,cross_val_score
from sklearn.metrics import classification_report, confusion_matrix
from sklearn.neighbors import KNeighborsClassifier

Implementing Logistical Regression

In [36]: regression = LogisticRegression(solver='newton-cg',multi_class='multinom
ial')

In [37]: regression.fit(X_train,y_train)

/opt/anaconda3/lib/python3.7/site-packages/sklearn/utils/optimize.py:20
3: ConvergenceWarning: newton-cg failed to converge. Increase the numbe
r of iterations.
"number of iterations.", ConvergenceWarning)

Out[37]: LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=

True,
intercept_scaling=1, l1_ratio=None, max_iter=100,
multi_class='multinomial', n_jobs=None, penalty='l
2',
random_state=None, solver='newton-cg', tol=0.0001, v
erbose=0,
warm_start=False)

In [38]: pred = regression.predict(X_test)

In [39]: regression.score(X_test,y_test)

Out[39]: 0.697495183044316

Our accuracy score for logistical regression is not completely ideal.

Implementing KNN Classification

In [40]: knn = KNeighborsClassifier(n_jobs=-1)

In [41]: knn.fit(X_train,y_train)
pred = knn.predict(X_test)
knn.score(X_test,y_test)

Out[41]: 0.6608863198458574

localhost:8888/nbconvert/html/Desktop/Python Data Products for Predictive Anlaytics Specialization/Course 2 - Design Thinking and Predictive Analytics for Data P… 4/5
8/21/2020 Course 2 - Final Project

In [43]: # We see that our accuracy scores for KNN Classifier is somewhat similar

In [44]: print(classification_report(y_test,pred))

precision recall f1-score support

0 0.42 0.32 0.37 118

1 0.20 0.37 0.26 19
2 0.76 0.83 0.79 358
3 0.00 0.00 0.00 24

accuracy 0.66 519

macro avg 0.35 0.38 0.35 519
weighted avg 0.63 0.66 0.64 519

In [45]: # We see above a detailed report of our KNN classification that provides
precision, recall, f-1 and a support scores
# We can observe a good f1 score here

Conclusion
Logistical Regression provided us a better accuracy than knn classifiers.

In [ ]:

localhost:8888/nbconvert/html/Desktop/Python Data Products for Predictive Anlaytics Specialization/Course 2 - Design Thinking and Predictive Analytics for Data P… 5/5