model_configuration2 {MSML} | R Documentation |
model_configuration2 function
Description
This function is similar to the model_configuration function, with the added capability to maintain constant variables across models during training and prediction (see cov_train and cov_valid in page 2). Additionally, users have the option to select between linear or logistic regression models.
Usage
model_configuration2(
data_train,
data_valid,
mv,
cov_train,
cov_valid,
model = "lm"
)
Arguments
data_train |
This includes the dataframe of the training dataset in a matrix format |
data_valid |
This includes the dataframe of the validation dataset in a matrix format |
mv |
The total number of columns in data_train/data_valid |
cov_train |
This includes dataframe of covariates for training dataset in a matrix format |
cov_valid |
This includes dataframe of covariates for validation dataset in a matrix format |
model |
This is the type of model (e.g. lm (default) or glm (logistic regression)) |
Value
This function will generate all possible model outcomes for validation and test dataset
Examples
data_train <- data_train
data_valid <- data_valid
mv=8
cov_train <- cov_train
cov_valid <- cov_valid
out=model_configuration2(data_train,data_valid,mv,cov_train, cov_valid, model = "lm")
#This process will produce predicted values for the validation datasets,
#corresponding to each model configuration trained on the training dataset.
#The outcome of this function will yield variables named 'predict_validation'
#and 'total_model_configurations.
#To print the outcomes run out$predict_validation and out$total_model_configurations.
#For details (see https://github.com/mommy003/MSML).
#If a user intends to employ logistic regression without constant covariates,
#we advise preparing a covariate file where all values are set to 1.