Changelog

Source: NEWS.md

causalKNN 0.3

  • Added direct causal KNN estimation and a wrapper function causalKNN_direct for off-the-shelf estimation. The back end bootstrapIndexMatrixProcessorTest function added for estimating direct causal KNN treatment effects for the test set. The package documentation and vignette has been updated to reflect new functionality

  • Eliminated unneeded parameters in previous functions and renamed tep_projection_enet to tep_enet for documentation consistency

  • Added bootstrap functionality for the Optimal K and causal KNN treatment effect estimation procedures. A bootstrap vignette has been added

causalKNN 0.2

  • Added Separate K among untreated NN and treated NN functionality with knn_optimal_k_separate

  • Added a wrapper function causalKNN_TEP that wraps the keys functions from the last commit and provides an off-the-shelf estimation

  • Added Travis-CI build support

  • Added Unit Test functionality with the testthat package

    • Added a test to ensure causalKNN_TEP yields the same results as running the sequential functions
    • Added tests to ensure the separate KNN approach works and yields the same results when K is set to be equal for treated and untreated

causalKNN 0.1

  • Initialized package using devtools and documentation is built with roxygen2

  • Added key functions causalknn_treatment_effect, knn_index_mat, knn_optimal_k, tep_projection_enet for a four step sequential estimation of the causal KNN regression and then a treatment effect projection with the Elastic Net

  • Added helper functions bootstrapIndexMatrixProcessor, standardizeVector, bootstrapKNNStepProcessor, cumWeightedSum, and cumWeightedSumStep. These functions are documented with @keyword internal in accordance to roxygen2 so they are still directly accessible to the user but hidden from manual

  • Added a the causalKNN-Intro vignette