iterateModelTT: Train-test approach for model tuning

In saharaja/imputeORS: Impute Missing Values in the BLS Occupational Requirements Survey

Description

Function to do train-test modeling in an iterative fashion, for the purposes of model tuning. The first prediction (Iteration 0) is just the naive guess. All subsequent iterations rely on XGBoost to produce preliminary predictions. These predictions are then adjusted to adhere to boundary conditions on the data (all estimates must be in the range [0,1], and the sum of all estimates within an occupational group must be <=1).

Usage

iterateModelTT(
  ors.data,
  n.iter,
  weight.step,
  mdl.d = 14,
  mdl.n = 200,
  mdl.e = 0.6,
  fold.list = NULL
)

Arguments

ors.data	Original data augmented with relevant predictors, i.e. all records, including both known and missing estimates (output of setDefaultModelingWeights(), or computeSimulations())
n.iter	Number of times to iterate/adjust the model
weight.step	Increment by which to increase modeling weight of test set data with each iteration
mdl.d	Tree model maximum depth; default is 14
mdl.n	Tree model rounds; default is 200
mdl.e	Tree model eta; default is 0.6
fold.list	A list of preset folds, if these have to be fixed across runs; default is to generate a new set of folds

Value

A list of length two, containing a list of hold out indices for the test set, and the results of iterative modeling

See Also

setDefaultModelingWeights()

computeSimulations()

saharaja/imputeORS documentation built on Feb. 4, 2022, 12:27 a.m.