Machine Learning For Diabetes

• uci machine learning diabetes dataset used for predicition

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

diabetes=pd.read_csv("diabetes.csv")
print(diabetes.shape)

(768, 9)

diabetes.head()

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	Pregnancies	Glucose	BloodPressure	SkinThickness	Insulin	BMI	DiabetesPedigreeFunction	Age	Outcome
0	6	148	72	35	0	33.6	0.627	50	1
1	1	85	66	29	0	26.6	0.351	31	0
2	8	183	64	0	0	23.3	0.672	32	1
3	1	89	66	23	94	28.1	0.167	21	0
4	0	137	40	35	168	43.1	2.288	33	1

Diabetes prediciton is classification task

• check databalance which is quite good

diabetes.Outcome.value_counts()

0    500
1    268
Name: Outcome, dtype: int64

• ohhh! 34% adults suffer from diabetes

import seaborn as sns
sns.countplot(diabetes['Outcome'],label="Count")

<matplotlib.axes._subplots.AxesSubplot at 0x2db63f3d828>

diabetes.isna().sum()

Pregnancies                 0
Glucose                     0
BloodPressure               0
SkinThickness               0
Insulin                     0
BMI                         0
DiabetesPedigreeFunction    0
Age                         0
Outcome                     0
dtype: int64

diabetes.dtypes

Pregnancies                   int64
Glucose                       int64
BloodPressure                 int64
SkinThickness                 int64
Insulin                       int64
BMI                         float64
DiabetesPedigreeFunction    float64
Age                           int64
Outcome                       int64
dtype: object

diabetes.columns.tolist()

['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age',
 'Outcome']

• check outlier and gaussian shape

diabetes[['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']].hist(figsize=(16, 10), bins=50, xlabelsize=8, ylabelsize=8);

outlier cleaning

• Pregnancies more than 10 is ideally not good so we consider it as outlier
• Body mass index is weight to height ration so weight less than 12 is not range of adults so we consider it as outlier
• bloodpressure lower than 40 is criticly low pressure so we consider it as outlier
• Glucose lower than 40 is criticly low pressure so we consider it as outlier
• SkinThickness lower than 60 is criticly low pressure so we consider it as outlier

removed all outlier

diabetes=diabetes[diabetes['Pregnancies']<=11]

diabetes=diabetes[diabetes['BMI']>=12]

diabetes=diabetes[diabetes['BloodPressure']>40]

diabetes=diabetes[diabetes['Glucose']>40]

diabetes=diabetes[diabetes['SkinThickness']<60]

diabetes[['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']].hist(figsize=(16, 10), bins=50, xlabelsize=8, ylabelsize=8);

Correlation

• after removal of outlier you can check distribution is likely normal
• we are going to use correlation for finding independent variable correlation

Assumption For PCC

• data should be normalize so we make it by normalizer
• linear you can check some features to be linear
• normal distributed
• applied always on continues variables

from sklearn.preprocessing import Normalizer
normalized_application = Normalizer().fit_transform(diabetes[['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']])
#print (normalized_application)

normal_df=pd.DataFrame(normalized_application)
normal_df.columns=['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']

cor=normal_df.corr()
plt.figure(figsize=(8,8))
flatui = ["#9b59b6", "#3498db", "#95a5a6", "#e74c3c", "#34495e", "#2ecc71"]
sns.heatmap(cor, vmax=1, square=True,annot=True,cmap=flatui)
plt.title('Correlation between different fearures')

Text(0.5,1,'Correlation between different fearures')

PCC Interpretation

• Pregnancies and age correlated which is right
• Glucose and BMI,age correlated which is right
• Blood pressure related to Glucose,BMI and Age which gives instinct about diabetes

as per the interpreation we are going to retain all independet variables for model

colnames=['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']
sns.pairplot(cor[colnames],size=1.5,x_vars=colnames,y_vars=colnames)

plt.show()

Check linearity in above graph which is good sign of correlation

diabetes.isna().sum()

Pregnancies                 0
Glucose                     0
BloodPressure               0
SkinThickness               0
Insulin                     0
BMI                         0
DiabetesPedigreeFunction    0
Age                         0
Outcome                     0
dtype: int64

normal_df.isna().sum()

Pregnancies                 0
Glucose                     0
BloodPressure               0
SkinThickness               0
Insulin                     0
BMI                         0
DiabetesPedigreeFunction    0
Age                         0
dtype: int64

print(diabetes.shape)
print(normal_df.shape)

(694, 9)
(694, 8)

diatbetes=diabetes.reset_index(drop=True)

normal_df=normal_df.reset_index(drop=True)

normal_df['Outcome'] = diabetes.Outcome.values

from sklearn.model_selection import train_test_split
df_train, df_val = train_test_split(normal_df, test_size=0.30)

print(df_train.shape)
print(df_val.shape)

(485, 9)
(209, 9)

Logistic classification model with 0.62 Accuracy

• with good algorithm we can achieve more accuracy

from sklearn.linear_model import LogisticRegression
features = normal_df.drop(["Outcome"], axis=1).columns

logreg = LogisticRegression().fit(df_train[features], df_train['Outcome'])



print("Training set score: {:.3f}".format(logreg.score(df_train[features], df_train['Outcome'])))
print("Validation set score: {:.3f}".format(logreg.score(df_val[features], df_val['Outcome'])))

Training set score: 0.682
Validation set score: 0.622

testdiabetes=pd.read_csv("testdiabetes.csv")
print(testdiabetes.shape)

(1, 8)

Test Data Prediction

normalized = Normalizer().fit_transform(testdiabetes)
#print (normalized_application)

normal_test=pd.DataFrame(normalized)
normal_test.columns=['Pregnancies',
 'Glucose',
 'BloodPressure',
 'SkinThickness',
 'Insulin',
 'BMI',
 'DiabetesPedigreeFunction',
 'Age']


prediction=logreg.predict(normal_test)

print(prediction)

[0]