Package 'TangledFeatures'

Title: Feature Selection in Highly Correlated Spaces
Description: Feature selection algorithm that extracts features in highly correlated spaces. The extracted features are meant to be fed into simple explainable models such as linear or logistic regressions. The package is useful in the field of explainable modelling as a way to understand variable behavior.
Authors: Allen Sunny [aut, cre]
Maintainer: Allen Sunny <[email protected]>
License: MIT + file LICENSE
Version: 0.1.1
Built: 2025-02-25 06:23:11 UTC
Source: https://github.com/allen-1242/tangledfeatures

Help Index

Automatic Data Cleaning

Description

Automatic Data Cleaning

Usage

DataCleaning(Data, Y_var)

Arguments

Data

The imported Data Frame
Y_var

The X variable

Value

The cleaned data.

Examples

DataCleaning(Data = TangledFeatures::Housing_Prices_dataset,  Y_var = 'SalePrice')

Generalized Correlation function

Description

Generalized Correlation function

Usage

GeneralCor(df, cor1 = "pearson", cor2 = "polychoric", cor3 = "spearman")

Arguments

df

The imported Data Frame
cor1

The correlation metric between two continuous features. Defaults to pearson
cor2

The correlation metric between one categorical feature and one cont feature. Defaults to biserial
cor3

The correlation metric between two categorical features. Defaults to Cramers-V

Value

Returns a correlation matrix containing the correlation values between the features

Examples

GeneralCor(df = TangledFeatures::Advertisement)

Housing prices dataset

Description

Housing prices dataset

The main TangledFeatures function

Description

The main TangledFeatures function

Usage

TangledFeatures(
  Data,
  Y_var,
  Focus_variables = list(),
  corr_cutoff = 0.85,
  RF_coverage = 0.95,
  plot = FALSE,
  fast_calculation = FALSE,
  cor1 = "pearson",
  cor2 = "polychoric",
  cor3 = "spearman"
)

Arguments

Data

The imported Data Frame
Y_var

The dependent variable
Focus_variables

The list of variables that you wish to give a certain bias to in the correlation matrix
corr_cutoff

The correlation cutoff variable. Defaults to 0.8
RF_coverage

The Random Forest coverage of explainable. Defaults to 95 percent
plot

Return if plotting is to be done. Binary True or False
fast_calculation

Returns variable list without many Random Forest iterations by simply picking a variable from a correlated group
cor1

The correlation metric between two continuous features. Defaults to pearson correlation
cor2

The correlation metric between one categorical feature and one continuous feature. Defaults to bi serial correlation correlation
cor3

The correlation metric between two categorical features. Defaults to Cramer's V.

Value

Returns a list of variables that are ready for future modelling, along with other metrics

Examples

TangledFeatures(Data = TangledFeatures::Advertisement, Y_var = 'Sales')