Final report Peer review

[amanjain454]

Team analysis citi bike ride data and tries to model the expected ride length.

Good things:

1. Results discuss feature importance, which makes the model more interpretable.
2. Interesting visualizations.
3. Overlaying weather data is an interesting idea.

Improvement areas:

1. I felt a general lack of clarity on the objective. Bike theft, social bias, probability of ending locations, bike available for redistribution is mentioned as objective/application at different places. These might be interrelated, but I think its too much promise or trying to solve too many problems from one model output.
2. Weather data granularity should be at least at hour level at least. Overlaying rainfall on day bike was out is leaving good scope for error.
3. Modeling (classifier for trips more than one hour) does not really match with stated objective (predicting trip duration)