computeConvergence: Compute convergence iteration of k-folds CV models

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

Description

For a set of iterative k-folds modeling results, it is necessary to determine the convergence point of the model. This function computes this using the RMSE of iterated predictions vs. the actual known values (this can be accomplished because the k-folds procedure creates mock missing data from the set of known estimates, so their true values are known quantities).

Usage

computeConvergence(test.fold.data, verbose = TRUE, show.plot = TRUE)

Arguments

test.fold.data	Test fold predictions (output of getTestFoldData())
verbose	Should messages be printed; default is TRUE (print messages)
show.plot	Should plot of RMSE by iteration be displayed; default is TRUE (display plot)

Details

We defined convergence as when the difference between the RMSE of consecutive iterations was < 0.001. If this convergence criteria is not met, the function simply returns the final iteration (and a message stating that convergence was not reached). Recall that Iteration 0 is simply the output of smart guessing (no modeling involved).

Value

Convergence iteration of k-folds cross validation modeling

See Also

iterateModelKFCV()

getTestFoldData()

ModelMetrics::rmse()

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