02 — HDD Regression & Normalization

Run OLS regression per postal code, apply quality filters, normalize slopes by building stock characteristics from the tax roll, and produce a thermal intensity metric for ranking.

2.1 — Load Data

2.2 — Per-Postal-Code OLS Regression

2.3 — Regression Quality Checks

2.4 — Merge with MPAC Data and Compute Heated Volume

2.5 — Normalize: Per-Property Slope and Thermal Intensity

2.6 — Effect of Normalization

Compare raw slope ranking vs. normalized ranking — they should differ substantially.

2.7 — Thermal Intensity by Structure Type

2.8 — Validate Against Ground Truth

Because this is synthetic data, we can check if the normalized metric correlates with the true slope.

2.9 — Save Results

---

Next: Notebook 03 ranks postal codes, identifies priority neighbourhoods, and builds the targeting output.