Back to portfolio GitHub Repo
// project_03 · Exploratory Data Analysis

Toronto Housing Market
5-Year EDA (2020–2024)

A 12-chart visual narrative surfacing what city-wide aggregate reports miss — neighbourhood-level divergence, the real spring premium, the affordability threshold crossing, and why days-on-market predicted the 2022 correction two months before prices moved.

COVID Freeze
Mar – May 2020
Volume -40%. Prices briefly dipped. Condo demand collapsed as WFH began.
Frenzy
Jun 2020 – Mar 2022
All-time price records. DOM under 10 days. 107%+ over asking bids.
Rate Shock
Apr – Dec 2022
BoC hiked 425 bps. Detached fell 20% peak-to-trough in 9 months.
New Normal
2023 – 2024
Tight supply, persistent affordability crisis, slow rate-cut recovery.
Pythonpandasmatplotlib seabornJupyterscipy
// price_trajectory

5-Year Price Trajectory

Average Sale Price — All Property Types (Monthly, 2020–2024)

Avg across 20 neighbourhoods × 4 property types · Phase shading: red=COVID, amber=Frenzy, purple=Rate shock, green=New Normal

+39%
Jan 2020 → Peak
From ~$850K to ~$1.18M in just 26 months. The fastest sustained appreciation in Toronto's recorded history.
-20%
Peak → Trough
Mar 2022 → Dec 2022. The sharpest 9-month correction since 2008. Detached led the decline; condos fell ~12%.
+25%
2020 → 2024 net
Despite the 2022 correction, the average home in Toronto ended 2024 25% above its Jan 2020 price.
// patterns_and_neighbourhoods

Seasonality & Neighbourhood Breakdown

Spring Premium — Consistent Every Year

Average seasonal index 2020–2024 · % above/below annual mean · Spring premium: +5–6%

Average Price by Neighbourhood (2024)

All property types averaged · Bridle Path excluded from chart for scale clarity

// affordability_crisis

The Affordability Threshold

Affordability Index by Property Type

Index = median household income ÷ min qualifying income · Below 1.0 = median household cannot qualify

Detached crossed below 0.5 in Q3 2021
The median Toronto household would need 2× their income to qualify for the median detached home. This threshold has not recovered even with 2024 rate cuts.
Condos remain the only accessible entry point
The condo affordability index stayed above 0.80 through most of the frenzy — the only property type a dual-income household could realistically qualify for at peak rates.
Rate cuts helped but didn't solve it
The BoC cut from 5.0% to ~4.0% through 2024. Affordability improved ~8% from trough — but remains worse than any point before 2021 for detached homes.
// key_findings

5 Most Actionable Findings

1. Days-on-market predicted the correction 2 months early
DOM started rising in April 2022. Prices didn't visibly peak until June. Watching DOM weekly would have given advance warning that no aggregate price report was giving.
2. Spring premium is real — +5–6% every year without exception
April–May consistently outperforms the annual average price across all 5 years, all property types. Buyers who close in December–February pay below the annual mean in every single year studied.
3. Suburban WFH premium emerged and partially reversed
Brampton, Mississauga, and Markham appreciated ~10% faster than downtown in 2020–21. By 2024, return-to-office erased about half of that premium as commute time became relevant again.
4. The correction was uneven — not one market
Detached fell ~20% peak-to-trough. Condos fell ~12%. Premium neighbourhoods (Rosedale, Forest Hill) barely corrected. Suburban detached saw the steepest drops. City-wide averages masked all of this.
5. Leslieville & Junction: fastest $/sqft growth — gentrification signal
Price-per-sqft in these neighbourhoods grew 35%+ over 5 years — faster than Rosedale. The heatmap picked this up as an anomaly worth investigating further with demographic and transit data.
Bonus: IQR doubled between 2020 and 2022
The spread between the 25th and 75th percentile prices doubled during the frenzy. The boom did not lift all boats equally — it widened the gap between neighbourhoods and property types.
// explore_more

Full 12-Chart Notebook

The complete Jupyter notebook with all 12 charts, a data generator calibrated to real TRREB market dynamics, and a findings summary. Runs end-to-end without any external data download.

View on GitHub ← All Projects