easy_fit: This function performs the training of the chosen models to a...

In elpidiofilho/easyFit: Routines for fit classification and regression models in easy way

Description

This function performs the training of the chosen models to a group of outcomes (y)

Usage

easy_fit(dy, dx, nfolds = 10, repeats = NA, resample_ = "cv",
  metric = ifelse(is.factor(dy[, 1]), "Kappa", "Rsquared"), cpu_cores = 7,
  tune_length = 5, fun_RFE = caret::rfFuncs, repeats_RFE = 1,
  metric_RFE = ifelse(is.factor(dy[, 1]), "Kappa", "Rsquared"),
  nfolds_RFE = 5, sizes_RFE = c(2:10, 15), tolerance_RFE = 0,
  models = c("ridge", "rf", "cubist", "pls", "foba", "gbm", "glmboost"),
  preprocess = NULL, seeds = NULL, verbose = TRUE)

Arguments

dy	outcomes
dx	co-variables
nfolds	Number of folds to be build in crossvalidation
repeats	repeats
resample_	resample method 'boot', 'boot632', 'optimism_boot', 'boot_all', 'cv', 'repeatedcv', 'LOOCV', 'LGOCV','none', 'oob', 'timeslice', 'adaptive_cv', 'adaptive_boot', 'adaptive_LGOCV'
metric	metric used to evaluate model fit. For numeric outcome ("RMSE", "Rsquared)
cpu_cores	Number of CPU cores to be used in parallel processing
tune_length	By default, this argument is the number of levels for each tuning parameters that should be generated by train.
fun_RFE	A list of functions for model fitting, prediction and variable importance
repeats_RFE	repeats to recursive feature elimination
metric_RFE	metric to recursive feature elimination
nfolds_RFE	nfolds to recursive feature elimination
sizes_RFE	A numeric vector of integers corresponding to the number of features that should be retained
tolerance_RFE	tolerance
models	vector of models to be fitted. Uses algortims names from Caret package.
preprocess	pre process
seeds	seeds
verbose	show results and messages from processing functions

Details

details

Author(s)

Elpidio Filho, [e-mail]

Examples

## Not run: 
easy_fit(dy,dx,train,models = c("ridge", "rf", "cubist"), metric = "Rsquared", tolerance = 2)

## End(Not run)

elpidiofilho/easyFit documentation built on May 28, 2019, 8:36 p.m.