Changelog
Source:NEWS.md
CGMissingDataR 0.0.2
Major changes
run_missing_glucose_imputation()now handles both explicit missing glucose values and missing readings implied by timestamp gaps. When timestamps skip expected CGM intervals, the function regularizes each subject to the expected interval and imputes the newly created missing glucose rows.The returned data frame is now simpler. It contains the user’s original columns plus
imputed_glucose_value. Internal columns used for timestamp regularization, lag features, rolling means, model fitting, and missingness tracking are no longer returned.The original glucose column is still preserved. Values that were originally missing, or created from timestamp gaps, remain
NAin the original target column, while completed values are stored inimputed_glucose_value.imputed_glucose_valueis returned as a continuous numeric model estimate. Users who need whole-number glucose values for reporting can round this column after imputation.Added a bundled Shiny app for interactive missing glucose imputation. The app lets users upload a CSV file or load example data, choose the relevant columns, run imputation, preview results, and download the completed data.
Added built-in example data for demonstrating both explicit missing glucose values and timestamp-gap handling.
The optional Python-compatible backend remains available with
imputer_backend = "sklearn". The default backend remainsimputer_backend = "mice"for standard R usage.Updated README and vignettes to describe timestamp-gap handling, the simplified output structure, the bundled Shiny app, backend options, and post-imputation rounding.