7/5/23, 12:49 PM Stroke Prediction

Heart Stroke Prediction

Importing libraries
In [ ]: #Importing the libraries
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.svm import SVC
from sklearn.neighbors import KNeighborsClassifier
from sklearn import metrics
from sklearn.metrics import accuracy_score
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import f1_score
from sklearn.metrics import mean_squared_error
from sklearn.metrics import log_loss

In [ ]: #Loading the dataset

df = pd.read_csv('healthcare-dataset-stroke-data.csv')
df.head()

Out[ ]: id gender age hypertension heart_disease ever_married work_type Residenc

0 9046 Male 67.0 0 1 Yes Private

Self-
1 51676 Female 61.0 0 0 Yes
employed

2 31112 Male 80.0 0 1 Yes Private

3 60182 Female 49.0 0 0 Yes Private

Self-
4 1665 Female 79.0 1 0 Yes
employed

In [ ]: df.drop('id', axis=1, inplace=True)

Data Preprocessing
In [ ]: df.describe()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 1/14

7/5/23, 12:49 PM Stroke Prediction

Out[ ]: age hypertension heart_disease avg_glucose_level bmi s

count 5110.000000 5110.000000 5110.000000 5110.000000 4909.000000 5110.00

mean 43.226614 0.097456 0.054012 106.147677 28.893237 0.04

std 22.612647 0.296607 0.226063 45.283560 7.854067 0.2

min 0.080000 0.000000 0.000000 55.120000 10.300000 0.00

25% 25.000000 0.000000 0.000000 77.245000 23.500000 0.00

50% 45.000000 0.000000 0.000000 91.885000 28.100000 0.00

75% 61.000000 0.000000 0.000000 114.090000 33.100000 0.00

max 82.000000 1.000000 1.000000 271.740000 97.600000 1.00

In [ ]: df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5110 entries, 0 to 5109
Data columns (total 11 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 gender 5110 non-null object
1 age 5110 non-null float64
2 hypertension 5110 non-null int64
3 heart_disease 5110 non-null int64
4 ever_married 5110 non-null object
5 work_type 5110 non-null object
6 Residence_type 5110 non-null object
7 avg_glucose_level 5110 non-null float64
8 bmi 4909 non-null float64
9 smoking_status 5110 non-null object
10 stroke 5110 non-null int64
dtypes: float64(3), int64(3), object(5)
memory usage: 439.3+ KB

In [ ]: df['age'].astype(int)

Out[ ]: 0 67
1 61
2 80
3 49
4 79
..
5105 80
5106 81
5107 35
5108 51
5109 44
Name: age, Length: 5110, dtype: int32

In [ ]: #Checking for null values

df.isnull().sum()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 2/14

7/5/23, 12:49 PM Stroke Prediction

Out[ ]: gender 0
age 0
hypertension 0
heart_disease 0
ever_married 0
work_type 0
Residence_type 0
avg_glucose_level 0
bmi 201
smoking_status 0
stroke 0
dtype: int64

In [ ]: #replacing the missing values with the most frequent value

df['bmi'].fillna(df['bmi'].mode()[0], inplace=True)

Check values and their count in the columns

In [ ]: print(df['ever_married'].value_counts())
print(df['work_type'].value_counts())
print(df['gender'].value_counts())
print(df['Residence_type'].value_counts())
print(df['smoking_status'].value_counts())

ever_married
Yes 3353
No 1757
Name: count, dtype: int64
work_type
Private 2925
Self-employed 819
children 687
Govt_job 657
Never_worked 22
Name: count, dtype: int64
gender
Female 2994
Male 2115
Other 1
Name: count, dtype: int64
Residence_type
Urban 2596
Rural 2514
Name: count, dtype: int64
smoking_status
never smoked 1892
Unknown 1544
formerly smoked 885
smokes 789
Name: count, dtype: int64

Replacing the values in columns with numerical values

Residence Type: Urban = 1, Rural = 0

Smoking Status: formerly smoked = 1, never smoked = 2, smokes = 3, Unknown = 0
Ever_Maried : Yes = 1, No = 0
Gender : Male = 1, Female = 0, Other = 2

file:///E:/Data Science Course/Projects/Stroke Prediction.html 3/14

7/5/23, 12:49 PM Stroke Prediction

Work Type : Private = 0, Self-employed = 1, children = 2, Govt_job = 3,

Never_worked = 4

In [ ]: df['ever_married'].replace({'Yes':1, 'No':0}, inplace=True)

df['gender'].replace({'Male':1, 'Female':0,'Other':2}, inplace=True)
df['Residence_type'].replace({'Urban':1, 'Rural':0}, inplace=True)
df['smoking_status'].replace({'formerly smoked':0, 'never smoked':1, 'smokes':2,
df['work_type'].replace({'Private':0, 'Self-employed':1, 'children':2, 'Govt_job

Exploratory Data Analysis

Find correlation between the variables

In [ ]: df.corr()['stroke'][:-1].sort_values().plot(kind='bar')

Out[ ]: <Axes: >

In [ ]: plt.figure(figsize=(10,10))
sns.heatmap(df.corr(), annot=True)

Out[ ]: <Axes: >

file:///E:/Data Science Course/Projects/Stroke Prediction.html 4/14

7/5/23, 12:49 PM Stroke Prediction

In [ ]: # replace age with number wrt to age group

# 0 = 0-12 , 1 = 13-19 , 2 = 20-30 , 3 = 31-60 , 4 = 61-100
df['age'] = pd.cut(x=df['age'], bins=[0, 12, 19, 30, 60, 100], labels=[0, 1, 2,
df.head()

Out[ ]: gender age hypertension heart_disease ever_married work_type Residence_type

0 1 4 0 1 1 0 1

1 0 4 0 0 1 1 0

2 1 4 0 1 1 0 0

3 0 3 0 0 1 0 1

4 0 4 1 0 1 1 0

Visulaizing the data

In [ ]: sns.countplot(x = 'gender', data = df)

file:///E:/Data Science Course/Projects/Stroke Prediction.html 5/14

7/5/23, 12:49 PM Stroke Prediction

Out[ ]: <Axes: xlabel='gender', ylabel='count'>

In [ ]: fig, ax = plt.subplots(4,4,figsize=(20, 20))

sns.countplot(x = 'gender', data = df,hue = 'stroke', ax=ax[0,0])
sns.countplot(x = 'age', data = df,hue = 'hypertension', ax=ax[0,1])
sns.countplot(x = 'age', data = df,hue = 'heart_disease', ax=ax[0,2])
sns.countplot(x = 'age', data = df,hue = 'stroke', ax=ax[0,3])
sns.countplot(x = 'hypertension', data = df,hue = 'stroke', ax=ax[1,0])
sns.countplot(x = 'heart_disease', data = df,hue = 'stroke', ax=ax[1,1])
sns.countplot(x = 'ever_married', data = df,hue = 'stroke', ax=ax[1,2])
sns.countplot(x = 'age', data = df,hue = 'ever_married', ax=ax[1,3])
sns.countplot(x = 'work_type', data = df,hue = 'stroke', ax=ax[2,0])
sns.countplot(x = 'Residence_type', data = df,hue = 'stroke', ax=ax[2,1])
sns.countplot(x = 'smoking_status', data = df,hue = 'stroke', ax=ax[2,2])
sns.lineplot(x = 'bmi', y = 'avg_glucose_level', data = df,hue = 'stroke', ax=ax
sns.countplot(x = 'age', data = df,hue = 'smoking_status', ax=ax[3,0])
sns.countplot( x = 'work_type', data = df,hue = 'Residence_type', ax=ax[3,1])
sns.countplot(x = 'work_type', data = df,hue = 'smoking_status', ax=ax[3,2])
sns.countplot(x = 'Residence_type', data = df,hue = 'smoking_status', ax=ax[3,3]

Out[ ]: <Axes: xlabel='Residence_type', ylabel='count'>

file:///E:/Data Science Course/Projects/Stroke Prediction.html 6/14

7/5/23, 12:49 PM Stroke Prediction

Train Test Split

In [ ]: X_train, X_test, y_train, y_test = train_test_split(df.drop('stroke', axis=1), d

Model Training

Logistic Regression
In [ ]: lr = LogisticRegression()
lr

Out[ ]: ▾ LogisticRegression

LogisticRegression()

In [ ]: #training the model

lr.fit(X_train, y_train)
lr.score(X_test, y_test)

file:///E:/Data Science Course/Projects/Stroke Prediction.html 7/14

7/5/23, 12:49 PM Stroke Prediction

C:\Users\DELL\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2k
fra8p0\LocalCache\local-packages\Python311\site-packages\sklearn\linear_model\_lo
gistic.py:458: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
n_iter_i = _check_optimize_result(
Out[ ]: 0.9393346379647749

In [ ]: #testing the model

lr_pred = lr.predict(X_test)
accuracy_score(y_test, lr_pred)

Out[ ]: 0.9393346379647749

Support Vector Machine (SVM)

In [ ]: from sklearn.svm import SVC
svm = SVC()
svm

Out[ ]: ▾ SVC

SVC()

In [ ]: #training the model

svm.fit(X_train, y_train)
svm.score(X_test, y_test)

Out[ ]: 0.9393346379647749

In [ ]: #testing the model

sv_pred = svm.predict(X_test)
accuracy_score(y_test, sv_pred)

Out[ ]: 0.9393346379647749

Decision Tree Classifier

In [ ]: from sklearn.tree import DecisionTreeClassifier
dt = DecisionTreeClassifier()
dt

Out[ ]: ▾ DecisionTreeClassifier

DecisionTreeClassifier()

In [ ]: #training the model

dt.fit(X_train, y_train)
dt.score(X_test, y_test)

file:///E:/Data Science Course/Projects/Stroke Prediction.html 8/14

7/5/23, 12:49 PM Stroke Prediction

Out[ ]: 0.9099804305283757

In [ ]: #testing the model

dt_pred = dt.predict(X_test)
accuracy_score(y_test, dt_pred)

Out[ ]: 0.9099804305283757

K-Nearest Neighbors (KNN)

In [ ]: knn = KNeighborsClassifier()
knn

Out[ ]: ▾ KNeighborsClassifier

KNeighborsClassifier()

In [ ]: #training the model

knn.fit(X_train, y_train)
knn.score(X_test, y_test)

Out[ ]: 0.9373776908023483

In [ ]: #testing the model

knn_pred = knn.predict(X_test)
accuracy_score(y_test, knn_pred)

Out[ ]: 0.9373776908023483

Model Evaluation

Logistic Regression
In [ ]: sns.heatmap(metrics.confusion_matrix(y_test, lr_pred), annot=True, fmt='d')
plt.title('Accuracy Score: {}'.format(accuracy_score(y_test, lr_pred)))
plt.ylabel('Predicted')
plt.xlabel('Actual')
plt.show()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 9/14

7/5/23, 12:49 PM Stroke Prediction

In [ ]: print('Logistic Regression Model Accuracy Score:',accuracy_score(y_test, lr_pred

print('Logistic Regression Model F1 score: ',metrics.f1_score(y_test, lr_pred))
print('Logistic Regression Model Mean Absolute Error: ',metrics.mean_absolute_er
print('Logistic Regression Model Mean Squared Error: ',metrics.mean_squared_erro
print('Logistic Regression Model log loss: ',log_loss(y_test, lr_pred))

Logistic Regression Model Accuracy Score: 0.9393346379647749

Logistic Regression Model F1 score: 0.0
Logistic Regression Model Mean Absolute Error: 0.060665362035225046
Logistic Regression Model Mean Squared Error: 0.060665362035225046
Logistic Regression Model log loss: 2.1866012819229583

Support Vector Machine (SVM)

In [ ]: sns.heatmap(metrics.confusion_matrix(y_test, sv_pred), annot=True, fmt='d')
plt.title('Accuracy Score: {}'.format(accuracy_score(y_test, sv_pred)))
plt.ylabel('Predicted')
plt.xlabel('Actual')
plt.show()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 10/14

7/5/23, 12:49 PM Stroke Prediction

In [ ]: print('SVM Model Accuracy Score:',accuracy_score(y_test, sv_pred))

print('SVM Model F1 score: ',metrics.f1_score(y_test, sv_pred))
print('SVM Model Mean Absolute Error: ',metrics.mean_absolute_error(y_test, sv_p
print('SVM Model Mean Squared Error: ',metrics.mean_squared_error(y_test, sv_pre
print('SVM Model log loss: ',log_loss(y_test, sv_pred))

SVM Model Accuracy Score: 0.9393346379647749

SVM Model F1 score: 0.0
SVM Model Mean Absolute Error: 0.060665362035225046
SVM Model Mean Squared Error: 0.060665362035225046
SVM Model log loss: 2.1866012819229583

Decision Tree Classifier

In [ ]: sns.heatmap(metrics.confusion_matrix(y_test, dt_pred), annot=True, fmt='d')
plt.title('Accuracy Score: {}'.format(accuracy_score(y_test, dt_pred)))
plt.ylabel('Predicted')
plt.xlabel('Actual')
plt.show()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 11/14

7/5/23, 12:49 PM Stroke Prediction

In [ ]: print('Decision Tree Model Accuracy Score:',accuracy_score(y_test, dt_pred))

print('Decision Tree Model F1 score: ',metrics.f1_score(y_test, dt_pred))
print('Decision Tree Model Mean Absolute Error: ',metrics.mean_absolute_error(y_
print('Decision Tree Model Mean Squared Error: ',metrics.mean_squared_error(y_te
print('Decision Tree Model log loss: ',log_loss(y_test, dt_pred))

Decision Tree Model Accuracy Score: 0.9099804305283757

Decision Tree Model F1 score: 0.2459016393442623
Decision Tree Model Mean Absolute Error: 0.09001956947162426
Decision Tree Model Mean Squared Error: 0.09001956947162426
Decision Tree Model log loss: 3.2446341602727773

K-Nearest Neighbors (KNN)

In [ ]: sns.heatmap(metrics.confusion_matrix(y_test, knn_pred), annot=True, fmt='d')
plt.title('Accuracy Score: {}'.format(accuracy_score(y_test, knn_pred)))
plt.ylabel('Predicted')
plt.xlabel('Actual')
plt.show()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 12/14

7/5/23, 12:49 PM Stroke Prediction

In [ ]: print('KNN Model Accuracy Score:',accuracy_score(y_test, knn_pred))

print('KNN Model F1 score: ',metrics.f1_score(y_test, knn_pred))
print('KNN Model Mean Absolute Error: ',metrics.mean_absolute_error(y_test, knn_
print('KNN Model Mean Squared Error: ',metrics.mean_squared_error(y_test, knn_pr
print('KNN Model log loss: ',log_loss(y_test, knn_pred))

KNN Model Accuracy Score: 0.9373776908023483

KNN Model F1 score: 0.0
KNN Model Mean Absolute Error: 0.06262230919765166
KNN Model Mean Squared Error: 0.06262230919765166
KNN Model log loss: 2.2571368071462796

Model Comparison
In [ ]: models = ['Logistic Regression', 'SVM', 'Decision Tree', 'KNN']
accuracy = [accuracy_score(y_test, lr_pred), accuracy_score(y_test, sv_pred), ac
plt.figure(figsize=(10,5))
plt.bar(models, accuracy, color = 'Maroon', width = 0.4)
plt.xlabel('Models')
plt.ylabel('Accuracy')
plt.title('Model Accuracy')
plt.show()

file:///E:/Data Science Course/Projects/Stroke Prediction.html 13/14

7/5/23, 12:49 PM Stroke Prediction

Conclusion
The model accuracies of Logistic Regression, SVM and KNN are quite similar i.e. 93.8 %.
The accuracy of Decision Tree Classifier is 91.8 %. So, we can use any of these models to
predict the heart stroke.

According to the graphs age v/s hypertension, heart disease showing chances of stroke,
the number of person having a stroke shows dependece upon heart disease and
hypertension. But when we plot the graph of heart disease and hypertension against the
stroke, the persons with lower chances of hypertension and heart disease has increased
chances of stroke. This is a peculiar thing and needs to be investigated further. In
addition to that non somkers have higher chances of stroke than smokers. This is also a
peculiar thing and needs to be investigated further. However person having BMI
between20 to 50 have higher chances of stroke.

Last but not least other features such as martial status, residence type as well as work
type are showing effect on the chances of stroke.

file:///E:/Data Science Course/Projects/Stroke Prediction.html 14/14