Use cross-validation to find the optimal nrounds for an Mixgb imputer. Note that this method relies on the complete cases of a dataset to obtain the optimal nrounds.
Usage
mixgb_cv(
data,
nfold = 5,
nrounds = 100,
early_stopping_rounds = 10,
response = NULL,
select_features = NULL,
xgb.params = list(),
stringsAsFactors = FALSE,
verbose = TRUE,
...
)Arguments
- data
A data.frame or a data.table with missing values.
- nfold
The number of subsamples which are randomly partitioned and of equal size. Default: 5
- nrounds
The max number of iterations in XGBoost training. Default: 100
- early_stopping_rounds
An integer value
k. Training will stop if the validation performance has not improved forkrounds.- response
The name or the column index of a response variable. Default:
NULL(Randomly select an incomplete variable).- select_features
The names or the indices of selected features. Default:
NULL(Select all the other variables in the dataset).- xgb.params
A list of XGBoost parameters. For more details, please check XGBoost documentation on parameters.
- stringsAsFactors
A logical value indicating whether all character vectors in the dataset should be converted to factors.
- verbose
A logical value. Whether to print out cross-validation results during the process.
- ...
Extra arguments to be passed to XGBoost.
Examples
params <- list(max_depth = 3, subsample = 0.7, nthread = 2)
cv.results <- mixgb_cv(data = nhanes3, xgb.params = params)
#> Multiple eval metrics are present. Will use test_rmse for early stopping.
#> Will train until test_rmse hasn't improved in 10 rounds.
#>
#> [1] train-rmse:1.963320±0.031171 test-rmse:2.021989±0.166801
#> [2] train-rmse:1.635703±0.036784 test-rmse:1.722371±0.175585
#> [3] train-rmse:1.425266±0.040358 test-rmse:1.541440±0.170586
#> [4] train-rmse:1.296524±0.040932 test-rmse:1.441356±0.174966
#> [5] train-rmse:1.211539±0.039615 test-rmse:1.371976±0.163681
#> [6] train-rmse:1.158323±0.039793 test-rmse:1.335619±0.149652
#> [7] train-rmse:1.119461±0.041829 test-rmse:1.318944±0.151686
#> [8] train-rmse:1.094165±0.038712 test-rmse:1.301223±0.155397
#> [9] train-rmse:1.074211±0.041667 test-rmse:1.289382±0.160836
#> [10] train-rmse:1.057512±0.040667 test-rmse:1.287639±0.176059
#> [11] train-rmse:1.044720±0.039432 test-rmse:1.286882±0.175661
#> [12] train-rmse:1.033794±0.039727 test-rmse:1.282698±0.181149
#> [13] train-rmse:1.022246±0.039490 test-rmse:1.276626±0.185726
#> [14] train-rmse:1.007901±0.035897 test-rmse:1.272276±0.179694
#> [15] train-rmse:1.000166±0.035963 test-rmse:1.271480±0.179027
#> [16] train-rmse:0.992511±0.037126 test-rmse:1.269869±0.180173
#> [17] train-rmse:0.985283±0.041051 test-rmse:1.272495±0.187991
#> [18] train-rmse:0.976042±0.036297 test-rmse:1.269337±0.182042
#> [19] train-rmse:0.968928±0.033149 test-rmse:1.270250±0.181051
#> [20] train-rmse:0.960499±0.031753 test-rmse:1.273157±0.179921
#> [21] train-rmse:0.954404±0.033598 test-rmse:1.268842±0.178302
#> [22] train-rmse:0.948135±0.034911 test-rmse:1.271862±0.177748
#> [23] train-rmse:0.939070±0.030973 test-rmse:1.271855±0.179913
#> [24] train-rmse:0.932000±0.033710 test-rmse:1.276054±0.172815
#> [25] train-rmse:0.925331±0.032511 test-rmse:1.280610±0.178959
#> [26] train-rmse:0.916091±0.031723 test-rmse:1.280791±0.177164
#> [27] train-rmse:0.910110±0.032802 test-rmse:1.287943±0.174522
#> [28] train-rmse:0.903795±0.032132 test-rmse:1.292798±0.175057
#> [29] train-rmse:0.898237±0.033322 test-rmse:1.298807±0.175802
#> [30] train-rmse:0.892281±0.034050 test-rmse:1.296191±0.181797
#> Stopping. Best iteration:
#> [31] train-rmse:0.883372±0.036310 test-rmse:1.303166±0.178872
#>
#> [31] train-rmse:0.883372±0.036310 test-rmse:1.303166±0.178872
cv.results$best.nrounds
#> [1] 21
imputed.data <- mixgb(
data = nhanes3, m = 3, xgb.params = params,
nrounds = cv.results$best.nrounds
)