model_evaluation {MSML} | R Documentation |
model_evaluation function
Description
This function will identify the best model in the validation and test dataset.
Usage
model_evaluation(dat, mv, tn, prev, pthreshold = 0.05, method = "R2ROC")
Arguments
dat |
This is the dataframe for all the combinations of the model in a matrix format |
mv |
The total number of columns in data_train/data_valid |
tn |
The total number of best models to be identified |
prev |
The prevalence of disease in the data |
pthreshold |
The significance p value threshold when comparing models (default 0.05) |
method |
The methods to be used to evaluate models (e.g. R2ROC (default) or r2redux) |
Value
This function will generate all possible model outcomes for validation and test dataset
Examples
dat <- predict_validation
mv=8
tn=15
prev=0.047
out=model_evaluation(dat,mv,tn,prev)
#This process will generate three output files.
#out$out_all, contains AUC, p values for AUC, R2, and p values for R2,
#respectively for all models.
#out$out_start, contains AUC, p values for AUC, R2, and p values for R2,
#respectively for top tn models.
#out$out_selected, contains AUC, p values for AUC, R2, and p values for R2,
#respectively for best models. This also includes selected features for models
#For details (see https://github.com/mommy003/MSML).
[Package MSML version 1.0.0.1 Index]