KPI Analyzer: Custom forecasting configuration in KPI mappings, allowing users to select forecasting model type and configure model-specific hyperparameters.
KPI Analyzer: Added ensemble model type for KPI models, which uses the model type configured on each KPI mapping, allowing users to mix different model types across predictors.
KPI Screener: Added support for hierarchical models, with KPIs returned in hierarchical order when sorting by company-level fields.
DSL: fs_revisions and va_revisions can now return error rates and confidence bands, and
kpi_model_revisions supports error rates relative to publish date. For details, see the
DSL reference documentation.
KPI Analyzer: New settings design with separate Global and Screen-specific sections for better organization of KPI Analyzer and Screener configuration.
KPI Analyzer: Simplified model sources popover with data category information, removed unnecessary nested popovers.
KPI Analyzer: Sources and weights popover now displays immediately while data category loads in the background.
KPI Mapper: Results page now uses the same modern UI and API endpoints as Data Catalog with improved coverage and predictive power summaries.
KPI Screener: Deleted watchlists are now clearly indicated and automatically excluded from evaluation.
KPI Screener: Screens now properly refresh when watchlists are updated.
Plotter: Charts automatically update when referenced watchlists are deleted.
Dashboards: Optimized widget rendering for dashboards with many charts.
Watchlists: Confirmation dialog now appears when clicking outside after editing watchlist titles.
Data Catalog: Removed irrelevant "KPI Overlapping" filter option when viewing single datasets.