City trip duration predictor
Inspired by publicly reported work at Uber
Predict how long a city taxi trip will take from pickup, drop-off and time.
The brief
Uber has written about DeepETA, the model it uses to predict arrival times. Start with the simpler version of the problem: using public city trip records, predict trip duration from pickup and drop-off points and the time of day, and see how far feature engineering takes a gradient-boosted model. Good looks like: RMSLE against a distance-and-speed baseline, maps of where errors are largest, and a short note on what a deep model would add.
Read the case first
What Uber has said in public about this kind of work, on its own pages:
- DeepETA: How Uber Predicts Arrival Times Using Deep Learning (Uber Blog) ↗
- Meet Michelangelo: Uber's Machine Learning Platform (Uber Blog) ↗
Make it your own
Bring it to an Indian city: autos, bike taxis or buses, using open ride data or trips you log yourself (no personal details). Show which features from New York don't carry over.
AITeky has no link with Uber, which has not reviewed or endorsed this brief. The brief only draws on what the company has published, uses open data rather than its data, and is a learning exercise: don't present your project as Uber's work or system.
What you will build
- Clean trips and remove impossible ones
- Compute distance, direction and time features
- Build a baseline from distance and average speed
- Train a gradient-boosted model
- Map errors by area and hour
Free, open data to build it with
Sign up free to get the data and links
It takes a minute with your email. AiTeky Pass (₹999/month or ₹9,999/year) is free for your first 30 days.