Project 21: Eclipse Cycles and Collective Mood

Objective

To investigate if Solar and Lunar eclipses correlate with measurable increases in collective distress or "chaos" in Seattle, WA.

Hypothesis

Physical Chaos: Seattle 911 call volumes (Fire/EMS) increase on the days of eclipses compared to normal days.

Data Sources

Methodology

911 Call Analysis (Detrending)

We employ a Multi-Factor Linear Regression model to strictly isolate "abnormal" volume from known temporal drivers.

  1. Feature Engineering: The following features are extracted for every day in the dataset:
    • Day of Week (Mon-Sun seasonality)
    • Month of Year (Annual seasonality)
    • Week of Month (Pay-cycle/monthly usage patterns)
    • Linear Time Index (Long-term population growth/trend)
  2. Regression: A Linear Regression model is trained on these features to predict the "Expected Call Count" for every day. No log transformation is applied to the count data; we analyze raw deviations.
  3. Residual Analysis: We calculate Actual Count - Expected Count = Residual.
  4. Normalization: Residuals are standardized into Z-Scores.
  5. Statistical Test: We perform two separate Independent T-Tests:
    • Solar Eclipses vs. Control Days
    • Lunar Eclipses vs. Control Days

Files

Usage

Run the analysis:

python3 analysis.py

## Data Provenance
### Eclipse Data
*   **Source**: NASA Eclipse Website.
*   **Link**: [https://eclipse.gsfc.nasa.gov/solar.html](https://eclipse.gsfc.nasa.gov/solar.html)
*   **Catalog**: Five Millennium Canon of Solar Eclipses.