Predicting blight in the city.
This repository contains files to accompany the capstone project "Predicting Blighted Buildings in Detroit."
Jupyter Notebook (Python) files:
- detroit_visualization.ipynb: code to visualize crimes, complaints, violations, and blight in Detroit in 2015
- modeling.ipynb: code for naive and full decision trees using count and/or MCM data
- vgg16.ipynb: image recognition using vgg16 as feature extractor
- xception.ipynb: image recognition using xception as feature extractor
- joined_model.ipynb: decision tree using all features plus result of image recognition
SQL files for PostgreSQL backend:
- neighborhoods.sql: load Detroit neighborhoods
- matches.sql: function to perform address matching
- crimes.sql: load crime incident data
- complaints.sql: load 311 complaint data
- violations.sql: load blight violations
- permits.sql: load demolition permit data
- parcels.sql: load parcel data and compute counts for modeling
JavaScript for interactive visualization:
- interface.html: JavaScript for interactive visualization
- parcels.json: Geojson file for interactive visualization
- neighborhoods.json: Geojson file for interactive visualization
For more information, refer to the project writeup available through Coursera.