Function to extract the feature weights from a SIAMCAT object

feature_weights(siamcat, verbose=1)

# S4 method for siamcat
feature_weights(siamcat, verbose = 1)

Arguments

siamcat

(Required). A siamcat-class object that contains trained models

verbose

integer, if the slot is empty, should a message be printed? values can be either 0 (no output) or 1 (print message)

Value

A dataframe containing mean/median feature weight and additional info or NULL

Details

The function extracts the weight matrix from all trained models (see weight_matrix) and computes several metrics on the feature weights:

  • mean.weight - mean weight across trained models

  • median.weight - median weight across trained models

  • sd.weight - standard deviation of the weight across trained models

  • mean.rel.weight - mean relative weight across trained models (each model is normalized by the absolute of all weights)

  • median.rel.weight - median relative weight across trained models

  • sd.rel.weight - standard deviation of the relative weight across trained models

  • percentage - percentage of models in which this feature was selected (i.e. non-zero)

Examples

data(siamcat_example)
temp <- feature_weights(siamcat_example)
#> Feature weights are empty
head(temp)
#> NULL