Save Philly By Data

This project is about saving Philadelphia by modeling crime, analyzing sustainability through tree locations, and running various machine learning models to understand particular trends. As a greater Philadelphia native myself, I care deeply about our city and aim to provide a fun web app that may help in some way. New features will be added. The model takes about 2 hours to update with new real-time data. If this message is visible, the project may still be initializing.

Crime Maps

  • 33 Different Crime Maps

    Generate crime maps based on 5-year intervals across 33 different crime categories

  • HTML

    Maps are viewed in the browser

  • Across Time

    Each marker represents a year of crime occurrence. See the legend to find out which color corresponds to each year.

Line Plots

Simple line graphs for all crimes

Linear Regression Plots

Across all months and years, sum particular crimes for that month and, over the entire interval, estimate M and run linear regression.

Linear Regression Animations (Source code in repo)

Linear regression from scratch with Matplotlib animations

Two-Sided Graphs

Two-sided graphs to highlight trends

Histograms or Distribution Plots

Distribution plots with fixed bandwidth (1.0), sampling the Gaussian distribution to handle zero cases

Additive Decomposition

Additive decomposition from seasonal decomposition, with a period of 30 to emphasize analyzing trends.

cProfile

cProfile execution graph to view runtime and identify potential improvements in the model.