常见机器学习模型的便捷使用（Python）

常见的机器学习模型有：逻辑回归（Logistic Regression，LR）、决策树（Decision Tree，DT）、随机梯度下降（Stochastic Gradient Descent，SGD）、支持向量机（Support Vector Machines，SVM）、随机森林（Random Forest，RF）、梯度提升决策树（Gradient Boosting Decision Tree，GBDT）、极限梯度提升（XGBoost）和LightGBM（Light Gradient Boosting Machine，LGBM）。
本文介绍了以上八种机器学习模型的便捷使用方法，实现了多个数据集对应多种方法。

1.导入所使用的包

from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.linear_model import SGDClassifier
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import GradientBoostingClassifier
from xgboost.sklearn import XGBClassifier
from lightgbm import LGBMClassifier
from sklearn.metrics import roc_auc_score
import warnings
warnings.filterwarnings("ignore")

2.定义模型（参数自行设置）

LR = LogisticRegression(random_state=100,tol=1e-6)  # 逻辑回归模型
DT = DecisionTreeClassifier(random_state=100) #决策树模型
SGD = SGDClassifier(random_state=100)  #随机梯度下降
SVM = SVC(probability=True,random_state=100,tol=1e-6)  # SVM模型
RF = RandomForestClassifier(n_estimators=100,random_state=100) #　随机森林
GBDT = GradientBoostingClassifier(random_state=100) #GBDT
XGB = XGBClassifier(random_state=100)  #XGBoost
LGBM = LGBMClassifier(random_state=100)  #LGBM

3.定义评价指标（AUC、SN、SP、ACC、MCC）

def muti_score(model,data_x,data_y,data_X,data_Y):
    model.fit(data_x,data_y)
    res = model.predict_proba(data_X)[:,1]
    auc = roc_auc_score(data_Y,res)
    print("AUC:",auc)
    pred = np.squeeze(res)
    f = pred>0.5
    pred[f]=1
    pred[pred<0.5]=0
    pred[pred==0.5]=0
    predict_label = pred
    pos_label = 1
    true_label = data_Y
    pos_num = np.sum(true_label==pos_label)
    print('pos_num=',pos_num)
    neg_num = true_label.shape[0]-pos_num
    print('neg_num=',neg_num)
    tp =np.sum((true_label==pos_label) & (predict_label==pos_label))
    print('tp=',tp)
    tn = np.sum(true_label==predict_label)-tp
    print('tn=',tn)
    sn = tp/pos_num
    sp = tn/neg_num
    acc = (tp+tn)/(pos_num+neg_num)
    fn = pos_num - tp
    fp = neg_num - tn
    print('fn=',fn)
    print('fp=',fp)
    tp = np.array(tp,dtype=np.float64)
    tn = np.array(tn,dtype=np.float64)
    fp = np.array(fp,dtype=np.float64)
    fn = np.array(fn,dtype=np.float64)
    mcc = (tp*tn-fp*fn)/(np.sqrt((tp+fn)*(tp+fp)*(tn+fp)*(tn+fn)))
    print(sn,sp,acc,mcc)

4.循环调用（可自行选择所用到的数据和模型）

M_train_name = ["train_x_1","train_x_2","train_x_3","train_x_4","train_x_5","train_x_6","train_x_7"]
M_test_name = ["test_x_1","test_x_2","test_x_3","test_x_4","test_x_5","test_x_6","test_x_7"]
model_name = ["LR","DT","SGD","SVM","RF","GBDT","XGB","LGBM"]  #选择已经定义好的模型
for i in range(len(M_train_name)):
    train_n = train_name[i]
    test_n = test_name[i]
    train_X = eval(train_n)
    test_X = eval(test_n)
    print(train_n,test_n)
    for name in model_name:
        model = eval(name)
        print(name)
        muti_score(model,train_X ,train_y,test_X,test_y)
        print("-"*50)  #分割不同模型的结果
    print("*"*100)  #分割不同数据的结果