Machine learning 如何评估XGBOOST概率模型？
下面是带有XGBOOST预测的验证表示例。在这里，我们试图预测一支球队在板球比赛的每一局结束时获胜的概率。数据以（局数、结束数、跑数/结束数、跑率）作为训练序列X和（目标）作为训练序列Y进行训练 目标变量由以下规则分配，即如果团队1=赢家，则1=其他0Machine learning 如何评估XGBOOST概率模型？,machine-learning,deep-learning,data-science,xgboost,supervised-learning,Machine Learning,Deep Learning,Data Science,Xgboost,Supervised Learning,下面是带有XGBOOST预测的验证表示例。在这里，我们试图预测一支球队在板球比赛的每一局结束时获胜的概率。数据以（局数、结束数、跑数/结束数、跑率）作为训练序列X和（目标）作为训练序列Y进行训练 目标变量由以下规则分配，即如果团队1=赢家，则1=其他0 |Match_Id|Inning|Over|Team 1|Team 2|winner|Bat_Team|Runs/Over|Inning Score|Run_Rate |Target|Pred | |--------|------|----|--
|Match_Id|Inning|Over|Team 1|Team 2|winner|Bat_Team|Runs/Over|Inning Score|Run_Rate |Target|Pred |
|--------|------|----|------|------|------|--------|---------|------------|---------|------|-----|
| a | 1 | 1 | A1 | A2 | A2 | A2 | 8 | 8 | 8.0 | 0 | 0.70|
| a | 1 | 2 | A1 | A2 | A2 | A2 | 16 | 24 | 12.0 | 0 | 0.82|
| a | 2 | 1 | A1 | A2 | A2 | A1 | 7 | 7 | 7.0 | 0 | 0.87|
| a | 2 | 2 | A1 | A2 | A2 | A1 | 5 | 12 | 6.0 | 0 | 0.95|
| b | 1 | 1 | A3 | A2 | A3 | A3 | 22 | 22 | 22.0 | 1 | 0.96|
| b | 1 | 2 | A3 | A2 | A3 | A3 | 16 | 38 | 19.0 | 1 | 0.82|
| b | 2 | 1 | A3 | A2 | A3 | A2 | 23 | 23 | 23.0 | 1 | 0.32|
| b | 2 | 2 | A3 | A2 | A3 | A2 | 12 | 35 | 17.5 | 1 | 0.90|
Import xgboost as xgb
def runXGB(train_X, train_y, seed_val=2):
param = {}
param['objective'] = 'binary:logistic'
param['eta'] = 0.05
param['max_depth'] = 8
param['silent'] = 1
param['eval_metric'] = "auc"
param['min_child_weight'] = 1
param['subsample'] = 0.7
param['colsample_bytree'] = 0.7
param['seed'] = seed_val
num_rounds = 100
plst = list(param.items())
xgtrain = xgb.DMatrix(train_X, label=train_y)
model = xgb.train(plst, xgtrain, num_rounds)
return model
model = runXGB(dev_X, dev_y)
xgtest = xgb.DMatrix(val_X)
preds = model.predict(xgtest)