[38]:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score
from sklearn.ensemble import RandomForestClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.tree import DecisionTreeClassifier
from sklearn.svm import SVC
from sklearn.preprocessing import LabelEncoder
​
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
####Read the stroke datafile####

[39]:

stroke_pred=pd.read_csv('/Users/lakuu2/Downloads/stroke_data/train_2.csv')
[40]:

stroke_pred.head()
[40]:
id	gender	age	hypertension	heart_disease	ever_married	work_type	Residence_type	avg_glucose_level	bmi	smoking_status	stroke
0	30669	Male	3.0	0	0	No	children	Rural	95.12	18.0	NaN	0
1	30468	Male	58.0	1	0	Yes	Private	Urban	87.96	39.2	never smoked	0
2	16523	Female	8.0	0	0	No	Private	Urban	110.89	17.6	NaN	0
3	56543	Female	70.0	0	0	Yes	Private	Rural	69.04	35.9	formerly smoked	0
4	46136	Male	14.0	0	0	No	Never_worked	Rural	161.28	19.1	NaN	0
####Checking for missing values####

[41]:

stroke_pred.isna().sum()
[41]:
id                       0
gender                   0
age                      0
hypertension             0
heart_disease            0
ever_married             0
work_type                0
Residence_type           0
avg_glucose_level        0
bmi                   1462
smoking_status       13292
stroke                   0
dtype: int64
####Replace "bmi" with its mean value####

[42]:

stroke_pred['bmi'].fillna(stroke_pred['bmi'].mean(), inplace=True)
[43]:

stroke_pred.isna().sum()
[43]:
id                       0
gender                   0
age                      0
hypertension             0
heart_disease            0
ever_married             0
work_type                0
Residence_type           0
avg_glucose_level        0
bmi                      0
smoking_status       13292
stroke                   0
dtype: int64
####Drop any missing values left####

[44]:

stroke_pred.dropna(inplace=True)
[45]:

stroke_pred.shape
[45]:
(30108, 12)
[46]:

stroke_pred.isna().sum()
[46]:
id                   0
gender               0
age                  0
hypertension         0
heart_disease        0
ever_married         0
work_type            0
Residence_type       0
avg_glucose_level    0
bmi                  0
smoking_status       0
stroke               0
dtype: int64
[47]:

stroke_pred.info()
<class 'pandas.core.frame.DataFrame'>
Int64Index: 30108 entries, 1 to 43399
Data columns (total 12 columns):
id                   30108 non-null int64
gender               30108 non-null object
age                  30108 non-null float64
hypertension         30108 non-null int64
heart_disease        30108 non-null int64
ever_married         30108 non-null object
work_type            30108 non-null object
Residence_type       30108 non-null object
avg_glucose_level    30108 non-null float64
bmi                  30108 non-null float64
smoking_status       30108 non-null object
stroke               30108 non-null int64
dtypes: float64(3), int64(4), object(5)
memory usage: 3.0+ MB
[48]:

stroke_pred.head(6)
[48]:
id	gender	age	hypertension	heart_disease	ever_married	work_type	Residence_type	avg_glucose_level	bmi	smoking_status	stroke
1	30468	Male	58.0	1	0	Yes	Private	Urban	87.96	39.2	never smoked	0
3	56543	Female	70.0	0	0	Yes	Private	Rural	69.04	35.9	formerly smoked	0
6	52800	Female	52.0	0	0	Yes	Private	Urban	77.59	17.7	formerly smoked	0
7	41413	Female	75.0	0	1	Yes	Self-employed	Rural	243.53	27.0	never smoked	0
8	15266	Female	32.0	0	0	Yes	Private	Rural	77.67	32.3	smokes	0
9	28674	Female	74.0	1	0	Yes	Self-employed	Urban	205.84	54.6	never smoked	0
####Perform label encoding to convert text values to numerical values####

[49]:

lb_make = LabelEncoder()
stroke_pred["gender"] = lb_make.fit_transform(stroke_pred["gender"])
stroke_pred["ever_married"] = lb_make.fit_transform(stroke_pred["ever_married"])
stroke_pred["work_type"] = lb_make.fit_transform(stroke_pred["work_type"])
stroke_pred["Residence_type"] = lb_make.fit_transform(stroke_pred["Residence_type"])
stroke_pred["smoking_status"] = lb_make.fit_transform(stroke_pred["smoking_status"])
[50]:

stroke_pred.head()
[50]:
id	gender	age	hypertension	heart_disease	ever_married	work_type	Residence_type	avg_glucose_level	bmi	smoking_status	stroke
1	30468	1	58.0	1	0	1	2	1	87.96	39.2	1	0
3	56543	0	70.0	0	0	1	2	0	69.04	35.9	0	0
6	52800	0	52.0	0	0	1	2	1	77.59	17.7	0	0
7	41413	0	75.0	0	1	1	3	0	243.53	27.0	1	0
8	15266	0	32.0	0	0	1	2	0	77.67	32.3	2	0
####Convert data type to integer or float####

[51]:

stroke_pred["gender"]= stroke_pred["gender"].astype(int)
stroke_pred["ever_married"]= stroke_pred["ever_married"].astype(int)
stroke_pred["work_type"]= stroke_pred["work_type"].astype(int)
stroke_pred["Residence_type"]= stroke_pred["Residence_type"].astype(int)
stroke_pred["smoking_status"]= stroke_pred["smoking_status"].astype(int)
####Perform Standard Scaler and make "stroke" variable the target variable####

[52]:

X = stroke_pred.drop(['stroke','id'], axis=1)
X=StandardScaler().fit_transform(X)
​
y = stroke_pred.stroke
/anaconda3/lib/python3.7/site-packages/sklearn/preprocessing/data.py:625: DataConversionWarning: Data with input dtype int64, float64 were all converted to float64 by StandardScaler.
  return self.partial_fit(X, y)
/anaconda3/lib/python3.7/site-packages/sklearn/base.py:462: DataConversionWarning: Data with input dtype int64, float64 were all converted to float64 by StandardScaler.
  return self.fit(X, **fit_params).transform(X)
####Split dataset into train and test data and set randon state####

[53]:

X_train, X_test, Y_train, Y_test = train_test_split(X,y, test_size=0.3, random_state=21)
[54]:

NB = GaussianNB()
DT = DecisionTreeClassifier()
RF = RandomForestClassifier()
SV = SVC(probability=True)
####Iterate through the models####

[55]:

Model_list=[NB, DT, RF, SV]
​
Accuracy=[]
​
for model in Model_list:
    score = cross_val_score(model, X_test, Y_test, cv=2)
    avg_score = np.mean(score)
    Accuracy.append(avg_score)
​
Model_name= ['NB', 'DT', 'RF', 'SV']
Model_performace= pd.DataFrame({'Model': Model_name, 'Accuracy': Accuracy})
Model_performace = Model_performace.sort_values(by='Accuracy', ascending=False)
Model_performace
[55]:
Model	Accuracy
3	SV	0.979741
2	RF	0.979520
1	DT	0.953061
0	NB	0.919185