xgboost介紹

2019-11-11 06:08:55

字體：大中小

供稿：網(wǎng)友

搞了一下午，終于把xgboost安上了，頭暈眼花，下面總結(jié)一下安裝過(guò)程和xgboost的原理因?yàn)椴幌刖幾g，在CSDN上找到編譯后的xgboost包網(wǎng)址為http://download.csdn.net/detail/onepiecehuiyu/9518553 在命令行中(win+R)cmd中進(jìn)入”…/xgboost-master/python-package/” 然后輸入”python setup.py install 解壓在”…/Lib/site-packages/xgboost-0.4-py3.5.egg”文件下的xgboost文件復(fù)制到”…/Lib/site-packages”下。 import xgboost as xgb無(wú)錯(cuò)誤，說(shuō)明安裝成功測(cè)試數(shù)據(jù)結(jié)果為[ 0.03688221]，性能在隨機(jī)森林和ExtraTrees之間

原理：

這里寫(xiě)圖片描述

測(cè)試代碼

# -*- coding: utf-8 -*-"""Created on Mon Feb 6 15:08:02 2017@author: Administrator"""import xgboost as xgbimport DTimport numpy as npTestSetN = 500dataset = DT.LoadData()y = []X = []for data in dataset: y += [data[-1]] X += [data[:-1]]dtrain = xgb.DMatrix(X[:-TestSetN], y[:-TestSetN])dtest = xgb.DMatrix(X[-1])evallist = [(dtest,'eval'), (dtrain,'train')]param = {'bst:max_depth':2, 'bst:eta':1, 'silent':1, 'objective':'reg:linear' }param['nthread'] = 4plst = param.items()num_round = 10bst = xgb.train(plst, dtrain)dtest = xgb.DMatrix(X[-TestSetN:-1] + [X[-1]])PRint((sum(np.transpose((bst.predict(dtest) - [y[-TestSetN:-1] + [y[-1]]])**2)/TestSetN)**0.5))

上一篇：substring用法總結(jié)

下一篇：算法Day12-層次遍歷二叉樹(shù)