compIndexBuilder 2.1.0

Data-format and missing-data update

• Added automatic recognition of wide indicator-year headers such as IN1-2019, IN1-2020, IN2-2019, and IN2-2020.
• Added automatic reshaping of that wide layout into panel form with one time column (Year) and one column per indicator (IN1, IN2, …), preserving the distinction between indicators and years.
• Added a selectable data-layout mode: Auto-detect, Standard/already tidy, or Force indicator-year wide format.
• Added configurable text missing-value codes. Common codes such as #N/A, N/A, NA, .., ..., and NULL can now be standardised to NA before numeric conversion.
• Added an explicit zero-as-missing option. Numeric 0 / 0.00 is preserved by default because zero may be a valid observation, and is converted to missing only when the user opts in.
• Added an import-guidance panel showing whether indicator-year columns were detected/reshaped, which years and indicators were recognised, and how many missing codes or zero values were converted.
• Preserved meaningful source headers instead of converting punctuation in names such as IN1-2019 before layout detection.
• Clarified the mixed-direction workflow: indicators for which high values are undesirable can be set once to “Lower is better” after reshaping.

compIndexBuilder 2.0.0

Major update

• Added workbook-wide multi-sheet Excel support.
• Added an active-sheet selector that lists all sheets in an uploaded workbook.
• Added controls to refresh the workbook sheet list and reload the active sheet.
• Added per-sheet CSV downloads and ZIP export of all workbook sheets.
• Unified CSV and Excel data-loading logic and removed duplicated server paths.
• Improved missing-data handling so missing indicators are not silently treated as zero.
• Added median, interpolation, and missForest-based imputation options.
• Added min-max and z-score normalisation and mixed indicator directions.
• Improved entity-level rankings for panel data and weight-impact comparisons.
• Improved single- and multi-entity time-series forecasting.
• Corrected time-period filtering in entity comparisons.
• Expanded pillar/sub-index construction with equal, custom, correlation-based, and PCA-based weighting.
• Added diagnostic outputs for Cronbach’s alpha, coefficient of variation, PCA, sensitivity analysis, correlation heatmaps, and Sankey diagrams.
• Added processed-data export.