12 · Personal & Academic Project
Ride-Share Demand Segmentation
Finding behavior patterns inside 693,000+ ride records
Business value
The analysis revealed distinct demand behaviors and achieved RMSE 8.36 for pricing prediction.
01 · Business problem
Ride prices varied widely, but there was no clear segmentation of user behavior or pricing drivers.
02 · My approach
I grouped rides by behavioral patterns first, then modeled pricing drivers within and across those segments.
03 · What I built
K-Means clustering, three demand tiers, and Random Forest regression across time, distance, and pricing conditions.
What changed my thinking
The corrections, tradeoffsand decisions behind it.
- 01
One average hid several behaviors
Segmentation showed that one-size-fits-all pricing analysis was misleading.
- 02
Clustering came before prediction
The groups provided context for interpreting pricing drivers.
- 03
Python only
The project was completed in Python without adding tools that were not actually used.