In JiaxiangBU/add2evaluation: add to evaluation of model

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Assume we train a xgboost model and want to save the params for further comparisons.

set.seed(123)
start_time <- now()
xgb_base05 <- xgb.train(
    data=dtrain03.0,
    ### 1
    eta = 0.3,
    nround=200,
    ### 2
    max_depth = 11,
    min_child_weight = 42,  
    gamma = 2.5,
    ### 3
    subsample = 0.8 ,
    colsample_bytree = 0.6,
    ### 评价标准
    ### eval.metric = "error",
    eval.metric = "mae",
    eval.metric = "rmse",
    ### eval.metric = ks_value,
    ### eval.metric = "auc",
    ### eval.metric = "logloss",
    ### objective
    objective = "reg:tweedie", ### 这是一个回归问题
    ### 其他
    seed = 123,
    watchlist=watchlist03.0,
    nfold = 30,
    early_stopping_rounds = 50,
    nthread = 8
    )
end_time <- now()
get_log_xgb(xgb_base05,start_time,end_time)

JiaxiangBU/add2evaluation documentation built on Jan. 21, 2020, 7:05 a.m.