sampleComplexityPAC: Sample Complexity based on PAC Learning Theory

In juscelino-izidoro/supcavs: Supervised and Unsupervised PCA Variable Selection

Description

This calculates the sample complexity for a range of possible models using PAC Learning concept with VC dimension.

Usage

sampleComplexityPAC(nVar, maxRem, error = 0.1, signif = 0.05)

Arguments

nVar	The number of variable of initial model.
maxRem	The number of variable to remove from the data set that will be used to compose the following models.
error	The error of PAC Learning. The range of this parameter is between 0 and 1.
signif	The significance of PAC Learning. This parameter ranges between 0 and 1.

Details

This function calculates the sample complexity for models (classifiers or regressions) that have "nVar", "nVar-1", "nVar-2", ..., "n_var-maxRem" input variables. This sample complexity is calculated using PAC Learning concept, considering some error and significance.

Value

The returned value is a vector with one sample complexity (as element) for each possible model.

Examples

  sampleComplexityPAC(10, 5)
  
  sampleComplexityPAC(30, 25, .05, .01)

juscelino-izidoro/supcavs documentation built on Jan. 2, 2022, 7:49 a.m.