I recently went through some online lectures about Bayesian methods, and found it to be useful to apply the concepts I've learned to a real-world application. NYC Open Data has a massive dataset on all taxi pickups, which we've parsed here to model taxi activity for each hour of the week across 288 regions of Manhattan.

I created three interactive visualizations from this project.

The first interactive shows mean hourly taxi pickups for a particular region over the course of a week.

The second interactive shows mean hourly taxi pickups across all regions of Manhattan for a given hour of the week.

The third interactive showcases the distribution of hourly pickups for a particular region at different times of the day.

The PDF doucment describes in detail the model I used and the derivations associated with implementing the Gibbs Sampler.

All my code is neatly organized in this iPython Notebook.

A summary of my entire project can be found in this post in my blog DataBucket

Always open to comments and suggestions for improvement!