随机森林优化预测氪金玩家开源

开源"/>

随机森林优化预测氪金玩家开源

        有道是最终结果的好坏，特征工程占大头，而模型和算法只占一小部分。但也需好的模型来进行优化，不敢妄称开源，代码写的还不规范，有问题一起交流，欢迎拍砖！

# 导入库文件
import pandas as pd
import numpy as np
from pandas import  Series,DataFrame
from sklearn.tree import DecisionTreeRegressorfrom sklearn.ensemble import BaggingRegressor
from sklearn.ensemble import RandomForestRegressor
from sklearn.ensemble import ExtraTreesRegressor# 第三类Boosting
from sklearn.ensemble import AdaBoostRegressor
from sklearn.ensemble import GradientBoostingRegressor# 数据预处理--读取数据并删除列:
def data_pre(train_name,test_name):print("开始读取....")train = pd.read_csv(str(train_name))test = pd.read_csv(str(test_name))train = train.drop('Unnamed: 0', axis=1)test = test.drop('Unnamed: 0.1', axis=1)test = test.drop('Unnamed: 0', axis=1)# 数据预处理--对数据集进行切分筛选拼接:train_X = train.iloc[:, 2:110]train_y = train.iloc[:, -2]train_X['multiple'] = train['multiple']# 先删除prediction price为0的行test = test.dropna(subset=['prediction_pay_price'])# 选出testx和testytest_X = test.iloc[:, 2:110]test_y = test.iloc[:, -2]test_X['multiple'] = test['multiple']print("读取完毕...")return train_X,train_y,test_X,test_y# 随机森林测试
# 第一类Bagging
def bagg_reg(trainx,trainy,testx,testy):# bag_clf = BaggingRegressor()bag_clf = BaggingRegressor()bag_clf.fit(trainx,trainy.astype('int'))score = bag_clf.score(testx, testy.astype('int'))return scoredef rand_reg(trainx,trainy,testx,testy):ran_clf = RandomForestRegressor()ran_clf.fit(trainx,trainy.astype('int'))score = ran_clf.score(testx, testy.astype('int'))return scoredef extra_reg(trainx,trainy,testx,testy):extra_clf = ExtraTreesRegressor()extra_clf.fit(trainx,trainy.astype('int'))score = extra_clf.score(testx, testy.astype('int'))return score# 第二类Boosting
def gb_reg(trainx,trainy,testx,testy):gb_clf = GradientBoostingRegressor(max_depth=2, n_estimators=30)gb_clf.fit(trainx,trainy.astype('int'))score = gb_clf.score(testx,testy.astype('int'))return score
def ada_reg(trainx,trainy,testx,testy):ada_clf = AdaBoostRegressor(DecisionTreeRegressor(max_depth=2), n_estimators=30)ada_clf.fit(trainx, trainy.astype('int'))score = ada_clf.score(testx, testy.astype('int'))return score
if __name__ == "__main__":trainX,trainy,testX,testy = data_pre("train_tap_high_test.csv","test_tap_high_test.csv")gb_score = gb_reg(trainX,trainy,testX,testy)ada_score = ada_reg(trainX, trainy, testX, testy)bagg_score = bagg_reg(trainX, trainy, testX, testy)ran_score = rand_reg(trainX, trainy, testX, testy)extra_score = extra_reg(trainX, trainy, testX, testy)print(extra_score)# print(score)