Changelog

Source: NEWS.md

nowcastNHSN 0.2.0

• nowcast_config() gains two opt-in controls for more robust probabilistic nowcasts. robust_sample = TRUE lets Skellam uncertainty sampling widen invalid variances instead of erroring when variance <= |pred|, and nonneg_pred = TRUE clamps negative predicted counts in final nowcast outputs to 0 (#34).
• run_single_nowcast() now passes robust_sample through to the configured uncertainty sampler and applies nonneg_pred after fitting.
• Documentation, examples, and tests were updated for the new configuration options.

nowcastNHSN 0.1.0

Initial release of nowcastNHSN, an R package providing tooling for nowcasting National Healthcare Safety Network (NHSN) reporting data.

Nowcast workflow

• nowcast_config() and nowcast_config_test() build and validate configuration objects for a nowcast run, with a dedicated print() method for inspection.
• run_single_nowcast() fits a single nowcast and run_state_nowcasts() fans the workflow out across U.S. states.
• summarize_nowcast() collapses posterior draws into the quantile summaries expected by downstream consumers (e.g. FluSight).
• read_nowcast_parquet() / write_nowcast_parquet() provide a parquet-based on-disk format for nowcast outputs.

Reporting-data sources

• fetch_reporting_data() is an S3 generic for pulling reporting data from a configurable source, with methods for:
  • cfa_dataops_source() — CFA DataOps backend (#30).
  • delphi_epidata_source() — Delphi epidatr API, with a lower-level fetch_reporting_data_epidatr() helper.
  • hub_data_source() — hubverse hubData archives (#24), including handling of FluSight reference dates via target_end_date (#27).
• fetch_nhsn_data() provides a convenience wrapper for the common NHSN pull.

Uncertainty utilities

• fit_normal() / fit_skellam() add parametric models for errors around point nowcasts and sample_normal() / sample_skellam() sample from them.
• get_uncertainty_fns() selects the matched fit/sample pair for use inside the nowcast pipeline.

Date and reporting helpers

• cumulative_to_incremental() converts cumulative reporting series to incremental counts.
• saturdays_to_epirange() maps a vector of Saturday dates to the epirange format used by epidatr.