机器学习中如何使参数最优

  

学习曲线 

from sklearn.datasets import load_breast_cancer
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import GridSearchCV
from sklearn.model_selection import cross_val_score
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

data = load_breast_cancer()
rfc = RandomForestClassifier(n_estimators=100,random_state=90)
score_pre = cross_val_score(rfc,data.data,data.target,cv=10,scoring='accuracy').mean()
print(score_pre)

scorel = []
for i in range(0,200,10):
    rfc = RandomForestClassifier(n_estimators=i+1,n_jobs=-1,random_state=90)
    score = cross_val_score(rfc,data.data,data.target,cv=10).mean()
    scorel.append(score)
print(max(scorel),(scorel.index(max(scorel))*10)+1)
plt.figure(figsize=[20,5])
plt.plot(range(1,201,10),scorel)
plt.show()

 

scorel = []
for i in range(65,75):
    rfc = RandomForestClassifier(n_estimators=i,n_jobs=-1,random_state=90)
    score = cross_val_score(rfc,data.data,data.target,cv=10).mean()
    scorel.append(score)
print(max(scorel),([*range(65,75)][scorel.index(max(scorel))]))
plt.figure(figsize=[20,5])
plt.plot(range(65,75),scorel)
plt.show()

网格搜索 对参数一个一个调整

data = load_breast_cancer()
param_grid = {'max_depth':np.arange(1,20,1)}
rfc = RandomForestClassifier(n_estimators=73,random_state=90)
GS = GridSearchCV(rfc,param_grid,cv=10)
GS.fit(data.data,data.target)
print(GS.best_params_)  # 显示调整出来最佳参数
print(GS.best_score_)    # 返回调整好的最佳参数对应的准确率

发现模型准确率下降了，推测在最佳泛化模型的左边，所以不调整max_depth，因为在模型左边，所以模型程度要往复杂调，所以对min_samples_leaf和min_samples_split不能做调整，所以只能调整max_features,来看模型如何变化。

data = load_breast_cancer()
param_grid = {'max_features':np.arange(5,30,1)}   
# max_feature 的默认值是最大特征数开平方
rfc = RandomForestClassifier(n_estimators=73,random_state=90)
GS = GridSearchCV(rfc,param_grid,cv=10)
GS.fit(data.data,data.target)
print(GS.best_params_)  # 显示调整出来最佳参数
print(GS.best_score_)    # 返回调整好的最佳参数对应的准确率

data = load_breast_cancer()
param_grid = {'criterion':['gini','entropy']}   
rfc = RandomForestClassifier(n_estimators=73,random_state=90)
GS = GridSearchCV(rfc,param_grid,cv=10)
GS.fit(data.data,data.target)
print(GS.best_params_)  # 显示调整出来最佳参数
print(GS.best_score_)    # 返回调整好的最佳参数对应的准确率

 发现怎么调准确率都会下降，说明模型已经在泛化误差最低点。