Cross-link OTIS (Ontario corrections) with TPS (Toronto police) data

R port of morie.otis_tps_overlay. Both feeds touch Toronto, so three overlay analyses are meaningful:

Details

• morie_otis_tps_yoy_correlation() – year-over-year Pearson r between OTIS Toronto-region segregation placements and TPS incident counts (per category).

• morie_otis_tps_per_region_rollup() – OTIS seg/RC totals per region x year, with the Toronto row flagged for overlay use.

• morie_otis_tps_composite_overlay() – alias for morie_otis_tps_yoy_correlation() (preserves the Python name, same body).

All three return a morie_otis_analysis_result (the shared RichResult-shaped list from otis_all_analyze.R).

Cross-year invariants

• OTIS UniqueIndividual_ID is reassigned every fiscal year (see variable_taxonomy.R); the overlay therefore joins at the year grain, never at the person grain.

• OTIS uses fiscal-year (EndFiscalYear); TPS uses calendar OCC_YEAR or REPORT_YEAR. The Pearson r here is computed on the year-aligned intersection – there is a small fiscal/calendar misalignment that is documented in the interpretation but not corrected. Toronto OTIS data covers only 2023-2025, so common-year samples are necessarily small.