San Francisco Fire Department Response Time and Priority Level Prediction

In the Stat Data Technologies (STA 141B) course at the University of California, Davis with Prof. Sharpnack, a final project was done utilizing a set of skills learned over the period of the quarter. A large public dataset from the San Francisco Fire Department was queried through their city’s web API. Webs scraping was done to create new features for the process of modeling. Various Data Visualization techniques were used to develop a greater insight into the dataset itself, all using Python’s available third-party packages. Different modeling techniques were also used to analyze various conceptual problems related to the data. Modeling included: Linear Regression, Logistic Regression, and Poisson Regression. In addition to the report itself, a 10-minute presentation was given discussing the progress of the project to the instructor. The presentation involved discussing the steps of the project within a Powerpoint. Link to project.

Plots from the project include (1) heatmap of emergency response calls by hour, (2) geoplot of the distribution of fire department response incidents, (3) comparison of spatial distribution of emergency incidents with Poisson distribution.