集成学习与Voting模型（员工离职率预测）

提供四种预测模型的方案和评估结果：KNN、逻辑回归、决策树分类、voting模型

# 1.导入数据，将特征和预测值进行分离，确定x,y值
import numpy as np 
import pandas as pd
worker = pd.read_csv('C:\\Users\\Liu\\Desktop\\data.csv')
x = worker.drop(columns = ['Attrition'])
y = worker['Attrition']

# 2.数据分割
from sklearn.model_selection import train_test_split
x_train,x_test,y_train,y_test = train_test_split(x,y,random_state = 666)

# 3.分别导入网格搜索调优的KNN,搜索调优的逻辑回归，搜索调优的决策树模型
from sklearn.neighbors import  KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier

#（1）网格搜索调优的KNN
knn_clf = KNeighborsClassifier()

from sklearn.model_selection import GridSearchCV
from sklearn.metrics import accuracy_score
param_knn = {
    'n_neighbors':[i for i in range(1,7)]
}
knn_grid = GridSearchCV(knn_clf,param_knn)
knn_grid.fit(x_train,y_train)
knn_grid.score(x_test,y_test)
knn_best_model = knn_grid.best_estimator_
knn_grid.best_score_
# predict_knn = knn_grid.predict(x_test)
# accuracy_score(y_test,predict_knn)

# (2)网格搜索调优的逻辑回归
lr_clf = LogisticRegression()

from sklearn.model_selection import GridSearchCV
param_lr = [
    {
       
        'C':[0.0001, 0.01, 0.1, 1],
        'penalty':['l1'],
        'solver':['liblinear']
    },
    {
        
        'C':[0.0001, 0.01, 0.1, 1],
        'penalty':['l2'],
        'solver':['newton-cg', 'lbfgs', 'liblinear', 'sag', 'saga']
    }  
]
       
lr_grid = GridSearchCV(lr_clf,param_lr)
lr_grid.fit(x_train,y_train)
lr_best_model = lr_grid.best_estimator_
lr_grid.best_score_

# (3)网格搜索调优的决策树模型
dt_clf = DecisionTreeClassifier(random_state=30,splitter='random',criterion = 'gini')
from sklearn.model_selection import GridSearchCV
param_dt = {
    'criterion':['gini', 'entropy'],
    'max_depth':[i for i in range(2,10)],
    'min_samples_leaf':[i for i in range(1,10)],
    'min_samples_split': [i for i in range(2, 10)]


}
dt_grid = GridSearchCV(dt_clf,param_grid=param_dt,cv = 8)
dt_grid.fit(x_train,y_train)
dt_grid.score(x_test,y_test)
dt_best_model = dt_grid.best_estimator_
dt_grid.best_score_

4.以上三个模型为基学习器，建立 voting 模型，并网格搜索调优
from sklearn.ensemble import VotingClassifier
voting_param = {
      'voting':['soft', 'hard']

}
voting_clf = VotingClassifier(
                estimators=[
                    ('knn_clf',knn_best_model),
                    ('lr_clf',lr_best_model),
                    ('dt_clf',dt_best_model )
                ]
)
model = GridSearchCV(voting_clf, cv=10, param_grid=voting_param)
model.fit(x_train,y_train)
model.best_score_