Ranked by how much each ward is above its own seasonal baseline β self-normalized so dense wards don't automatically win.
| Rank | Ward | Tier | Score | per 100k | vs. seasonal β |
|---|
Weekly city-wide activity, normalized to each year's mean. Baseline mean shows the typical shape; the current year is overlaid so you can see where 2026 is tracking.
Wildlife and bin-damage 311 calls by hour of day and day of week. Raccoons strike at night; Toronto reports them in the morning.
loadingβ¦
Each ward's weekly score is a weighted sum of 311 service requests across
seven raccoon-adjacent categories. Higher weights on literal wildlife calls
(Cadaver - Wildlife, Injured - Wildlife) and lower
weights on proxy signals (bin-lid repairs, missed garbage pickup, illegal dumping,
overflowing litter bins).
The vs. seasonal column is the ward's current score divided by its 2019 + 2022β2024 average for this ISO week. 2020β2021 are excluded because pandemic-era 311 patterns were badly non-representative. Values above 1.0 mean the ward is more active than usual for this time of year; below 1.0 means quieter. Each ratio carries a 90% credible interval derived from the Poisson noise in each category's weekly count.
The current-week projection on the home page blends a 4-week trailing mean with a weather adjustment fit on Toronto City daily temperature and precipitation (n β 200 baseline-year weeks, Ξ²_temp β 0.027 log-ratio per Β°C, Ξ²_precip β 0.001 per mm, RΒ² β 0.17). Temperature is the dominant signal; precipitation adds almost nothing after controlling for it, because in Toronto warm weeks tend to be wet weeks. Weather is observed, not forecast β we only use temps through the current week.
None of these records are literally raccoon records. The index is a composite of open data signals that raccoons would plausibly drive. 311 data reflects who calls in, not what actually happens β tree-lined neighbourhoods with more civically engaged residents will appear more active.
Empirical check on that: we tested whether ravine adjacency (per the City's Ravine & Natural Feature Protection polygons) predicts higher activity β the ravine-system-as-raccoon-highway hypothesis. It doesn't. Per 1000 residents, ravine-heavy wards report essentially the same number of wildlife calls as low-ravine wards (low/mid/high tertile means: 40, 44, 42 calls per 1000 residents over the baseline window). Toronto-Danforth, with little ravine area, tops the per-resident list anyway. Ravine geography does not meaningfully predict 311 wildlife reporting.
The ward leaderboard is ranked by seasonal ratio, not raw score, so dense downtown wards don't automatically dominate. The Score column is the raw weighted Index value; per 100k divides it by the ward's 2021 census population so wards of different sizes are directly comparable. Seasonal ratio self-normalizes too, since each ward is compared to its own historical baseline.
Source: 311 Service Requests β Customer Initiated, open.toronto.ca. Ward boundaries: City Wards. Ward populations: Ward Profiles (25-Ward Model), 2021 census. Ravine adjacency: Ravine & Natural Feature Protection Area. Weather: Environment and Climate Change Canada, Toronto City station. Inspired by the retired Toronto Raccoon Activity Index (April 1 joke).
Prior art on the same 311 data: SafePassage by Invisible Carbon pipes wildlife-call hotspots into real-time driver alerts for Waze and Google Maps β a safety / prevention angle on the same records this project uses. raccoonindex takes a different angle: a weekly civic-alert-style composite that combines wildlife calls with bin-damage, missed garbage, and illegal dumping as raccoon-proxy signals.