NEWS.md

In decorrelate: Decorrelation Projection Scalable to High Dimensional Data

decorrelate 0.1.6

• July 9, 2025
• submit to CRAN

decorrelate 0.1.5

• April 2, 2025
• update docs
• update sumInverseCorr()

decorrelate 0.1.4

• March 20, 2025
• update docs
• update sumInverseCorr() based on direction

decorrelate 0.1.3

• March 11, 2025
• in eclairs(), if svd() fails fall back on irlba()
• sumInverseCorr() has upper bound of p
• fix docs

decorrelate 0.1.2

• April 23, 2024
• use dmult() instead of transposing
• estimate nu to give correlation matrix close to having diagonals 1

decorrelate 0.1.1

• Feb 27, 2024
• update all functions to process covariance by shrinking correlation and retaining variance information

decorrelate 0.1.0

• Feb 27, 2024
• update all functions to process covariance by shrinking correlation and retaining variance information

decorrelate 0.0.19

• Jan 02, 2024
• add mahalanobisDistance()

decorrelate 0.0.18

• Sept 18, 2023
• add averageCorr(), averageCorrSq() and sumInverseCorr()
• add tests for these functions
• add alpha parameter to quadForm()

decorrelate 0.0.17

• add n.samples argument to \code{eclairs()}

decorrelate 0.0.16

• add eclairs_sq

decorrelate 0.0.15

• use irlba for SVD instead of PRIMME
• improve documentation

decorrelate 0.0.14

• add series_start_total() and use it in estimate_lambda_eb() for partial SVD
• this substantially improves estimation of lambda when partial SVD is used
• add averageCorr()

decorrelate 0.0.13

• add redundancy index as x.ri, y.ri
• add effective variance and effective dependency metrics

decorrelate 0.0.12

• fastcca() and cca() give equivalent results
• but canonical variates are rotated with respect to each other

decorrelate 0.0.11

• Scale Cxy by sqrt(1-lambda.x)*sqrt(1-lambda.y)
• Add Cramer's V statistic

decorrelate 0.0.10

• Sept 13, 2021
• add `fastcca() and cca()

decorrelate 0.0.9

• Sept 7, 2021
• add kappa() to compute condition number
• add logDet() to compute log determinant
• add cca() for canonical correlation analysis
• simplify examples

decorrelate 0.0.8

• July 3-12, 2021
• getCov() and getCor() now have lambda argument
• plot() for eclairs shows arrow on right for zero eigen-values
• estimate_lambda_eb() now returns logML for estimated or specified lambda
• this is stored by eclairs()

decorrelate 0.0.7

• June 1, 2021
• add whiten() that combines eclairs() and decorrelate() into one function call
• add eclairs_corMat() to perform decomposition on correlation matrix
• extend reform_decomp2() to work with result of eclairs_corMat()

decorrelate 0.0.6

• May 25, 2021
• add estimate_lambda_eb() to perform empirical Bayes estimation of lambda
• works for truncated SVD by setting missing ev's to the mean residual variance
• this makes estimation of lambda stable for varying ranks
• add plot() for eclairs

decorrelate 0.0.5

• add reform_decomp()
• improve documentation

decorrelate 0.0.4

• add lm_eclairs() and lm_each_eclairs()
• add R/lm_projection.R

decorrelate 0.0.1

• initial version

decorrelate documentation built on Aug. 8, 2025, 7:55 p.m.