A project to visualize house prices and affordabilty by neighborhoods and local regions.
Please note that not all the code used to generate this data and provide this experience is included. Many Jupyter notebooks and datasets were too large to include on Github. If you are interested in the methodology, please contact ethanghunt.
Thanks you to Mapbox for enabling this demo and continued development of their responsive mapping tool.
Get a Mapbox token, and replace MAPBOX_ACCESS_TOKEN_HERE in demo/public/js/mapbox.js on line 1.
Create a python virtual environment of your choice. Activate it. Ensure you are in the demo directory.
pip install -r requirements.txt
python router.py
Navigate to localhost:8000.
Handles exploratory data analysis and creation of model types.
Uses Ridge and Lasso regression models to perform informed variable selection, and then saves the final model as a pickle file. If you would like to load this model, please use the joblib package as shown in model.py.
In addendum, this handles the merging of dispersed data files including merging based off of nearest longitude, latitude pairs.
Contains nearly all scraped files and data collection required for the reproducability. This includes compressed zillow scrape data, official neighborhood boundaries, the official Atlanta city boundary, and multitudes of region-specific datasets. In addition, it handles the scraping from the Google API.
Other scraping methods are not detailed here since they lacked fully automatic processes, felt exploitive, or were not available by the time of code compilation (deleted).