Data Scientist · New Blog
Turning complex data into decisions.
Latest Post
I built and compared five regression models for predicting home sale prices. Ridge regression worked best: adding more features only helps if you stop the model from over-relying on any one of them.
Case study
How I led the build of one analytics product from country-level files across Europe, and saved 4 days every reporting cycle.
Case study
Setting up a 20-person field team, collecting 4,500 customer records, and helping restore power to 1,500 households after 7 years of blackout.
Post
Using Foursquare venue data and k-means clustering to find the Toronto neighbourhood with the right mix of foot traffic and demand, but not too much coffee shop competition already.
Post
A full walkthrough of a crisis referral risk model: data quality, feature engineering, model comparison, and what a screening tool like this can and can't tell a clinical team.
Post
Two weeks from the end of maternity leave: on 'rotting' a little, nurturing a lot, and finally starting this blog. Plus a bit about how I became a data scientist.