select_by_RW: Select predictor games by relative weight of predictors

In Blockhead-yj/report.iqz: Make Potential Predction Report of IQUIZOO

Select predictor games by relative weight of predictors

Description

Usually users will have lots of iquizoo games test at once, however it is bad idea to put all the games into a model to predict academic score, which might cause the problem of overfitting or a decline of adjust r squared. So we choose a subset of iquizoo games to be placed into the model by drop the least import variable, according to relative analysis.

Usage

select_by_RW(data_wider, formula, n_tasks = 5, least_RW = NULL)

Arguments

data_wider	Data.frame, which each row contains all the observations of a single person.
formula	Formula, indicate which variables should be included in model. Or a lm object, the formula will be the call of model, ie. model$call.
n_tasks	Numerical, a selection standard indicate how many tasks preserve.
least_RW	Numerical, a selection standard indicate the least relative weight the final selected task should have.

Value

a lm object with additional attribution "rw", relative weight

Blockhead-yj/report.iqz documentation built on March 18, 2022, 5:30 a.m.