diabetes model prediction
"""
# @Time : 2020/9/6
# @Author : Jimou Chen
"""
from sklearn.linear_model import LogisticRegression
import pandas as pd
import matplotlib.pyplot as plt
import seaborn
import numpy as np
import missingno as msn
from sklearn.metrics import classification_report
from sklearn.model_selection import train_test_split
def label_distribution(data):
p = data.Outcome.value_counts().plot(kind='bar') # 使用柱狀圖畫出
plt.show()
# 可視化數據發布, 有些數據本不該為0的卻為0，其實是空的
p = seaborn.pairplot(data, hue='Outcome')
plt.show()
# 把空值的用柱狀圖畫出來
p = msn.bar(data)
plt.show()
def handle_data():
data = pd.read_csv('data/diabetes.csv')
# 查看標簽分布
print(data.Outcome.value_counts())
# 把葡萄糖，血壓，皮膚厚度，胰島素，身體質量指數中的0替換為nan
handle_col = ['Glucose', 'BloodPressure', 'SkinThickness', 'Insulin', 'BMI']
data[handle_col] = data[handle_col].replace(0, np.nan)
# 設定閥值
thresh_count = data.shape[0] * 0.8
# 若某一列數據缺失的數量超過20%就會被刪除
data = data.dropna(thresh=thresh_count, axis=1)
# 填充數據，得到新的數據集data
data['Glucose'] = data['Glucose'].fillna(data['Glucose'].mean())
data['BloodPressure'] = data['BloodPressure'].fillna(data['BloodPressure'].mean())
data['BMI'] = data['BMI'].fillna(data['BMI'].mean())
return data
if __name__ == '__main__':
new_data = handle_data()
label_distribution(new_data)
# 切分數據集
x_data = new_data.drop('Outcome', axis=1)
y_data = new_data.Outcome
x_train, x_test, y_train, y_test = train_test_split(x_data, y_data, test_size=0.3, stratify=y_data)
# 建模
model = LogisticRegression()
model.fit(x_train, y_train)
# 預測
pred = model.predict(x_test)
# 評估
print(classification_report(pred, y_test))
D:\Anaconda\Anaconda3\python.exe D:/Appication/PyCharm/Git/kaggle-project/DiabetesPrediction/diabetes_predict.py
0 500
1 268
Name: Outcome, dtype: int64
precision recall f1-score support
0 0.90 0.80 0.85 169
1 0.58 0.76 0.66 62
accuracy 0.79 231
macro avg 0.74 0.78 0.75 231
weighted avg 0.81 0.79 0.80 231
Process finished with exit code 0
