COVID-19 Case Counts - Trend or Reporting Noise

Research question

Which features reflect the broad reported-case wave, and which reflect the reporting calendar?

Compare day-to-day reporting variation with a defensible descriptive reported-case trend for Lombardia during 1 July–31 December 2020. Use official public tables, a compact fixed period, centered seven-day and 21-day smoothers, and explicit uncertainty about the observation process.

Data and comparison design

The comparison uses centered seven- and 21-day smoothers with the raw daily reports. Summarize weekday departures from the seven-day curve to distinguish recurrent reporting patterns from the slower epidemic trajectory. Centering uses observations on both sides of each date.

Input or source	Role in this case	Source context
Italian Civil Protection pinned repository	Official regional CSVs, snapshot dated 2025-01-10 and retrieved 2026-09-29	Preserve upstream attribution and applicable source terms; this README grants no additional third-party rights
Regional field definitions	Use nuovi_positivi, not net active-case change; preserve notification dates	Preserve upstream attribution and applicable source terms; this README grants no additional third-party rights

Recorded findings

Reporting summary	Recorded result	Interpretation
Window	Lombardia, July–December 2020	Historical reported cases.
Period total	385,002	Reports, not reconstructed infections.
Smoothed peak	10 November for 7- and 21-day centered curves	Retrospective agreement across smoothing widths.
Weekday departures	Monday −44.6%; Friday +23.8%	Median deviations from the seven-day curve.

There are 184 target dates, 204 dates including boundary context, and 385,002 reported cases in the target period. The raw maximum is 11,489 on 7 November 2020. Both centered smoothers peak on 10 November: 8,806.285714 cases/day for seven days and 8,135.142857 for 21 days. Median daily departures from the seven-day smoother are −44.591495% on Mondays and +23.793565% on Fridays; the overall median absolute departure is 20.633769%. These source-count and smoothing calculations were independently checked in the bounded review.

What the comparison shows

Agreement on the peak date supports a retrospective description of the wave. Large weekday departures explain why individual daily reports can misrepresent its direction.

Assumptions and limits

The curves describe reported cases, not reconstructed infections. Centered windows use future observations and cannot serve as real-time nowcasts. Testing changes, delays, revisions and shared-window dependence limit interpretation; a later source snapshot does not reconstruct historical real-time data vintages.

Sources and research materials

The unchanged research package comes from the upstream archive. It retains the original conversation and included scientific artifacts, including analysis code, results and figures. The poster is an AI-generated thematic illustration, not study evidence. This introduction summarizes archived findings; the analysis was not rerun. Later bounded checks do not certify the entire study or mean the original package was corrected.