/AI/test_sklearn.py

import pandas as pd
import numpy as np
from sklearn.tree import DecisionTreeClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.metrics import confusion_matrix
from sklearn.metrics import accuracy_score, precision_score, recall_score



df_origin = pd.read_csv('heart_failure_clinical_records_dataset.csv')

df = df_origin
def reset_creatinine_phosphokinase(x):
    if x>1000:
        return round(11 + x/1000)
    return round(x/100)
df['creatinine_phosphokinase'] = df['creatinine_phosphokinase'].apply(lambda x: reset_creatinine_phosphokinase(x))
df['creatinine_phosphokinase']=df['creatinine_phosphokinase'].astype(int);
df['time']=df['time']/100
df['time']=df['time'].astype(int)

df['ejection_fraction']=df['ejection_fraction']/20
df['ejection_fraction']=df['ejection_fraction'].astype(int)


quantile = pd.Series(np.arange(4))
quantile[0] =df['platelets'].quantile(0.2)
quantile[1] =df['platelets'].quantile(0.4)
quantile[2] =df['platelets'].quantile(0.6)
quantile[3] =df['platelets'].quantile(0.8)
def reset_by_quantile(x, quant):
    if(x<quant[0]):
        return  0
    elif(x<quant[1]):
        return  1
    elif(x<quant[2]):
        return  2
    elif(x<quant[3]):
        return  3
    else:
        return 4

df['platelets']=df['platelets'].apply(lambda x: reset_by_quantile(x,quantile))
df['platelets']=df['platelets'].astype(int)

quantile = pd.Series(np.arange(4))
quantile[0] =df['serum_creatinine'].quantile(0.2)
quantile[1] =df['serum_creatinine'].quantile(0.4)
quantile[2] =df['serum_creatinine'].quantile(0.6)
quantile[3] =df['serum_creatinine'].quantile(0.8)
def reset_by_quantile(x, quant):
    if(x<quant[0]):
        return  0
    elif(x<quant[1]):
        return  1
    elif(x<quant[2]):
        return  2
    elif(x<quant[3]):
        return  3
    else:
        return 4

df['serum_creatinine']=df['serum_creatinine'].apply(lambda x: reset_by_quantile(x,quantile))
df['serum_creatinine']=df['serum_creatinine'].astype(int)

df = df.loc[:,['anaemia','creatinine_phosphokinase', \
               'serum_creatinine','ejection_fraction','diabetes','platelets', 'high_blood_pressure','sex', \
               'smoking','time','DEATH_EVENT']]
#realdata = df.loc[0:20,['age','anaemia','diabetes','high_blood_pressure','sex','smoking','DEATH_EVENT']]
#realdata['age']=round(realdata['age']/20) 
#realdata['age']=realdata['age'].astype(int);
#realdata = df.loc[0:20,['diabetes','sex','smoking','DEATH_EVENT']]

df_train = df.sample(frac=0.8)
rowlist=[]
for indexs in df_train.index:
	rowlist.append(indexs)
df_test=df.drop(rowlist,axis=0)

print(df.loc[0,:])
print(df.loc[1,:])
print(df.loc[150,:])
#print(df_train)
#print(df_test)

X = df_train.iloc[:,:10]
Y = df_train.iloc[:,10]

X_t = df_test.iloc[:,:10]
Y_t = df_test.iloc[:,10]
#print(X)
#print(Y)
#model = DecisionTreeClassifier()
model = GaussianNB()
model.fit(X,Y)
pred = model.predict(X_t)
print(pred)
print(Y_t)

confusion_matrix = confusion_matrix(Y_t,pred)

print(confusion_matrix)
accuracy = accuracy_score(Y_t,pred)
precision = precision_score(Y_t,pred)
recall = recall_score(Y_t,pred)
print(accuracy)
print(precision)
print(recall)