elder_weighted_corr_score: Calculate weighted corr score

In mdkessler/kTSCR: Implements k-Top Scoring Cluster Regression

Description

Calculates a weighted correlation score that over weights higher correlations. When calling elders, this allows me to prioritize features that have are a part of a smaller number of highly correlative (with y) pairwise features as opposed to features that are a part of a larger number of relative less correlative pairwise features.

Usage

elder_weighted_corr_score(corr, max_score = 100)

Arguments

corr	a float between 0 and 1 representing the absolute value of a correlation coefficient (r)
max_score	the maximum weight given to a correlation. Given that this weighted scoring can produce exponentially increasing weighted scores, setting this to 100 (representing a correlation of ~0.99) is reasonable.

Value

a float between 0 and 100, that represents a mapping of corr to a weighted score that overweights higher correlation values

Examples

corr <- 0.05
weighted_score <- elder_weighted_corr_score(corr)

mdkessler/kTSCR documentation built on Feb. 25, 2021, 10:31 p.m.