ML之LoR&DT&RF：基于LoR&DT(CART)&RF算法对mushrooms蘑菇数据集(22+1,6513+1611)训练来预测蘑菇是否毒性(二分类预测)

 

 

目录

输出结果

设计思路

核心代码

---

 

 

 

输出结果

0、数据集

after LabelEncoder

 

1、LoR算法

LoR_model_GSCV.grid_scores_： [mean: 0.77012, std: 0.01349, params: {'C': 0.001, 'penalty': 'l1'}, 
                              mean: 0.86936, std: 0.01035, params: {'C': 0.001, 'penalty': 'l2'}, 
                              mean: 0.91229, std: 0.01022, params: {'C': 0.01, 'penalty': 'l1'}, 
                              mean: 0.91045, std: 0.00831, params: {'C': 0.01, 'penalty': 'l2'}, 
                              mean: 0.94707, std: 0.00853, params: {'C': 0.1, 'penalty': 'l1'}, 
                              mean: 0.93599, std: 0.00841, params: {'C': 0.1, 'penalty': 'l2'}, 
                              mean: 0.95984, std: 0.00670, params: {'C': 1, 'penalty': 'l1'}, 
                              mean: 0.94953, std: 0.00790, params: {'C': 1, 'penalty': 'l2'}, 
                              mean: 0.96553, std: 0.00531, params: {'C': 10, 'penalty': 'l1'}, 
                              mean: 0.95722, std: 0.00559, params: {'C': 10, 'penalty': 'l2'}, 
                              mean: 0.96646, std: 0.00516, params: {'C': 100, 'penalty': 'l1'}, 
                              mean: 0.96599, std: 0.00528, params: {'C': 100, 'penalty': 'l2'}, 
                              mean: 0.96661, std: 0.00513, params: {'C': 1000, 'penalty': 'l1'}, 
                              mean: 0.96646, std: 0.00564, params: {'C': 1000, 'penalty': 'l2'}]
LoR_model_GSCV.best_score_： 0.96661024773042
LoR_model_GSCV.best_params_： {'C': 1000, 'penalty': 'l1'}
LoR_model_GSCV.best_score_： 0.96661024773042
LoR_model_GSCV.best_params_： {'C': 1000, 'penalty': 'l1'}
LoR_model_GSCV_auc_roc： 0.9739644970414202

2、DT算法

 

 

3、RF算法

 

RFC_model_GSCV grid_scores_： [mean: 0.99938, std: 0.00075, params: {'max_features': 'auto', 'min_samples_leaf': 10, 'n_estimators': 10}, 
                              mean: 0.99954, std: 0.00070, params: {'max_features': 'auto', 'min_samples_leaf': 10, 'n_estimators': 20},
                              …… 
                              mean: 0.97784, std: 0.01071, params: {'max_features': 'log2', 'min_samples_leaf': 80, 'n_estimators': 20}, 
                              mean: 0.98215, std: 0.00703, params: {'max_features': 'log2', 'min_samples_leaf': 80, 'n_estimators': 30}, 
                              mean: 0.98169, std: 0.00550, params: {'max_features': 'log2', 'min_samples_leaf': 90, 'n_estimators': 80}, 
                              mean: 0.98169, std: 0.00801, params: {'max_features': 'log2', 'min_samples_leaf': 90, 'n_estimators': 90}]
RFC_model_GSCV best_score_： 0.9998461301738729
RFC_model_GSCV best_params_： {'max_features': 'auto', 'min_samples_leaf': 10, 'n_estimators': 50}
RFC_model_GSCV_auc_roc： 1.0

 

设计思路

后期更新……

 

 

核心代码

后期更新……

RF 

tuned_parameters = {'min_samples_leaf': range(10,100,10), 
                       'n_estimators' : range(10,100,10),
                        'max_features': ['auto','sqrt','log2'] }
   
  
RFC_model_GSCV = GridSearchCV(RFC_model, tuned_parameters,cv=10)    
RFC_model_GSCV.fit(X_train,y_train)                                 
  
endtime = time.clock()
print ('RFC_model_GSCV Training time：',endtime - starttime)   

print('RFC_model_GSCV grid_scores_：', RFC_model_GSCV.grid_scores_)
print('RFC_model_GSCV best_score_：',  RFC_model_GSCV.best_score_)
print('RFC_model_GSCV best_params_：', RFC_model_GSCV.best_params_)

y_prob = RFC_model_GSCV.predict_proba(X_test)[:,1]   
y_pred = np.where(y_prob > 0.5, 1, 0)                
RFC_model_GSCV.score(X_test, y_pred)