feature.boruta.report: Create PDF and Text Reports About Selected Boruta Features

In MarcoNiemann/kaggle_house: Analysing House Prices in Ames, Iowa and Build a Sales Price Prediction Model

Description

This methods provides a PDF with the major info graphics related to Boruta feature selection, as well as a text file with all confirmed, tentative and rejected variables including the formula generated from them.

Usage

feature.boruta.report(boruta)

Arguments

boruta

The Boruta object for which a report should be generated.

Details

In a first step this method will generate two plots for the given Boruta object: One plot that will show boxplots for all features, as well as their importance and their selection status (e.g. a red feature has been discared, a green one confirmed). Additionally the imputation history is added, which shows the importance (and acceptance status) of each feature over the time. These two graphics are combined into one PDF file. Since it is versioned with time information, each call to this method will automatically generate a new report. The time stamp thus allows to version different Boruta objects. In case the Boruta object has been generated with feature.boruta.comp and contains the Confirmed (etc.) variables an additional text file is generated and filled with all the information regarding those features. If the formula switch has been activated the generated formula will also be added to that text file. Similar to the PDF file the txt file is versioned via timestamps as part of the file name.

Examples

 boruta <- KaggleHouse:::feature.boruta(
   target = data_train_na$SalePrice, predictors = data_train_na[-81],
    fixNA = T, roughFix = F, variables = T, selected = T, formula = T,
    verbose = T
 )
 KaggleHouse:::feature.boruta.report(boruta)

MarcoNiemann/kaggle_house documentation built on May 7, 2019, 2:50 p.m.