Edmonton Residential Property Value Analysis

My Role

Solo analyst. I independently scoped, cleaned, analyzed, and visualized a dataset of over 400,000 property records as a capstone project for M2M Tech's DataTalent program.

The Question

Edmonton's residential real estate market spans enormously diverse neighbourhoods. I wanted to understand: what factors correlate most strongly with property values, which neighbourhoods sit at the extremes, and how are properties distributed across price ranges?

This wasn't just a technical exercise — the same questions matter to city planners making development decisions, homeowners assessing their market position, and real estate professionals advising clients. Data-driven analysis at this level is what separates informed decisions from guesswork.

So What — A Product Lens

For a product team building tools for city planners, homeowners, or real estate professionals, an analysis like this shapes feature prioritization. A tool built for city planners needs ward-level and zoning aggregations. A tool for individual homeowners needs neighbourhood comparisons and percentile context. The same underlying data serves completely different user needs depending on who you're building for, which is why understanding your user segment before building your data product matters as much as the analysis itself.

Skills Demonstrated

Python · Pandas · NumPy · Bokeh · Data cleaning · Statistical analysis · Data visualization · Google Colab · Analytical storytelling

📎 Watch the recorded presentation → 📎 View the code on GitHub →

Approach.

Using Python in Google Colab with Pandas, NumPy, and Bokeh, I:

  • Cleaned and pre-processed the dataset (handled missing values, converted columns, removed outliers, and filtered for residential properties only)

  • Ran descriptive statistics and percentile analysis across the full dataset

  • Computed neighbourhood and ward-level summaries

  • Built 6 visualizations, including an interactive HTML visualization with hover tooltips and zoom

Code to import pandas, numpy, and bokeh
Code for cleaning the data, checking missing values, removing outliers and rows with missing assessed values

Basic Statistics and Percentiles.

10th

$15,500.00

25th

$185,000.00

50th/Median

$367,500.00

75th

$490,000.00

90th

$632,500.00

95th

$757,500.00

99th

$1,236,500.00

Mean

$370,925.80

Median

$367,500.00

Standard Deviation

$308,491.59

Minimum

$500.00

Maximum

$25,361,500.00

Neighbourhood Statistics.

Top 10 Most Expensive Neighbourhoods (by Mean Value)

A list of the top 10 most expensive neighbourhoods by mean value

Top 10 Least Expensive Neighbourhoods (by Mean Value)

Code to calculate the average assessed values by neighbourhood

Ward Statistics.

Most Expensive Ward (by mean value)

Pihêsiwin & Ipiihkoohkanipiaohtsi

Least Expensive Ward (by mean value)

O-day'min

A map of Edmonton's wards

Visualizations.

Bar chart of a distribution of property assessed values
A bar chart of the top 20 neighbourhoods by average assessed value
Bar chart of the top 5 wards by average assessed value
Geographic distribution of property values
Interactive map of Edmonton property values
A bar chart showing the correlation of each of the other values collected with assessed value

Key Findings.

Distribution Most residential properties fall below $500K, with a median assessed value of $367,500. This positions Edmonton as relatively affordable compared to other major Canadian cities.

Neighbourhood Extremes The most and least expensive neighbourhoods vary significantly — with high-value properties concentrated in the southwest and central areas, and lower-value properties more distributed across the city.

Ward Level Pihêsiwin & Ipiihkoohkanipiaohtsi had the highest mean property value; O-day'min had the lowest.

Correlation No strong correlation was found between assessed value and the other variables in this dataset — which is itself a meaningful finding. It points to factors outside this dataset (i.e. school proximity, crime rates, transit access, land use zoning) as more likely drivers of value.

What I'd Do Differently

I'd enrich the dataset with external variables, such as transit proximity, school ratings, walkability scores, to build a more complete picture of what actually drives property values. The current analysis identifies that the correlation is weak; a richer dataset would start to explain why.

Previous
Previous

NAT Account Pages