nlcor: Compute Nonlinear Correlation

View source: R/correlations.R

nlcorR Documentation

Compute Nonlinear Correlation

Description

Compute nonlinear correlation using adaptive spatial sampling.

Usage

nlcor(
  x,
  y,
  refine = NA,
  plt = T,
  line_thickness = 1,
  line_opacity = 1,
  chart_title = NA
)

Arguments

x

A numeric vector. NAs are not allowed. The length of the vector should be more than 10.

y

A numeric vector. NAs are not allowed. Length should be same as 'x'. The order of 'x' and 'y' are relevant. 'nlcor(x, y)“ is not equal to 'nlcor(y, x)'. Therefore, 'x' should be a causal and 'y' should be a dependent variable.

refine

Optional. If manually set, a small value, e.g., 0.01, increases the granularity of local correlation computation. The runtime is faster if 'refine' is manually set. Otherwise, the algorithm automatically finds the best refinement.

plt

Optional. Default value FALSE. Set TRUE to return ggplot2 object for the data correlation visualization.

line_thickness

Optional. Default 1. Thickness of the correlation lines. It is a float argument > 0.

line_opacity

Optional. Default 1, completely opaque. The opacity of the correlation lines. A float between 0-1. 0 is transparent.

Value

The output is a list containing the nonlinear correlation cor.estimate, adjusted.p.value, and cor.plot. cor.estimate is between 0 and 1 (a negative nonlinear correlation is undefined). The adjusted.p.value shows the statistical significance of the cor.estimate. If adjusted.p.value > 0.05, the nonlinear correlation estimate can be considered as noise (statistically not significant). If plt = T, ggplot2 object is return for plotting the identified local linear correlations in the data.

See Also

cor

Examples

library(nlcor)
library(ggplot2)
ncor <- nlcor(x1, y1)
ncor <- nlcor(x2, y2, plt = TRUE)
ncor <- nlcor(x3, y3, refine = 0.01, plt = TRUE)


ProcessMiner/nlcor documentation built on June 1, 2022, 2:49 p.m.