Changelog

Source: NEWS.md

BatchRegression 0.0.21

• April 28, 2026
• add computeResiduals()
  • faster and uses less memory with nb_deviance()

BatchRegression 0.0.20

• April 20, 2026
• glmFitResponses() accepts parameter chunkSize

BatchRegression 0.0.19

• March 18, 2026
• use MatrixGenerics instead of matrixStats

BatchRegression 0.0.18

• March 17, 2026
• add nReaders = 1 argument to glmFitResponses() and glmmFitResponses()

BatchRegression 0.0.17

• Feb 26, 2026
• add Satterthwaite DDF
• fix hypothesis testing

BatchRegression 0.0.16

• Feb 5, 2026
• faster varpart()
• 100x speed increase for var_predict_terms()

BatchRegression 0.0.15

• Feb 2, 2026
• add results()

BatchRegression 0.0.14

• Jan 17, 2026
• Update response filtering

BatchRegression 0.0.13

• Dec 22, 2025
• performance improvements
• add ridge regression in fastglmm and fastlmm
• in variance partitioning analysis, add faster approximation of baseline rate for count models

BatchRegression 0.0.12

• Nov 16, 2025
• update varpart()
• glmm* and glm* return offset

BatchRegression 0.0.11

• May 6, 2025
• add lmmFitFeatures() and lmmFitResponses()
• move some code to fastglmm

BatchRegression 0.0.10

• name change
• refactor for compatibility across ecosystem

BatchRegression 0.0.9

• Use Thread Building Blocks (TBB)
• add predict()
• handle singular design matrices, set results to NaN
• compute and return nZeroPrediction

BatchRegression 0.0.8

• Feb 20, 2025
• add lmRidge()
• *lmFit*() return objects and add coef(), show(), etc.
• add dev_resids() in C++ and residuals() in R
• for NB model in both R and fastLinReg
  • family = “nb” sets dispersion to 1
  • family = “nb:theta” estimates dispersion
• add simulate() to simulate responses from estimated model parameters
• residuals(fit, type="parametricBootstrap") to simulate responses, compute residuals, then transform residuals of observed data using empirical mean and variance to get z-scores

BatchRegression 0.0.7

• Feb 12, 2025
• add colnames to coef and se
• add name checks for lmFitFeatures() and glmFitFeatures()
• consolidate input checks

BatchRegression 0.0.6

• Feb 3, 2025
• toList() now handles empty list
• this allows lmFitFeatures() to handle GenomicDataStream that is empty

BatchRegression 0.0.5

• Dec 18, 2024
• add support for family functions (i.e. gaussian(), quasibinomial())
• add glmFitFeatures(...,fastApprox=TRUE) using initial GLM fit, and preprojection on the working response

BatchRegression 0.0.4

• Dec 16, 2024
• bug fixes and convergence criteria
• add family can be array of nb:x for varying x for glmFitResponses

BatchRegression 0.0.3

• Dec 12, 2024
• Add glmFitFeatures() and glmFitResponses()

BatchRegression 0.0.2

• Dec 4, 2024
• Add support for GLM and NB GLM