/cityfinder

CS 122 PROJECT - where to live?

Primary LanguagePLpgSQL

cityfinder

What is this?

CS 122 project: An application to recommend US cities to move to based on user preferences.

Requirements

  • You are using Python 2.7.
  • You are using a VirtualMachine where you will download requirements OR you have virtualenv
  • All Python packages required are within requirements.txt

Set Up

To get all required packages, run

pip install -r requirements.txt

Using cityfinder

cd to the /cityfinder dir (where the file manage.py is) and run

python manage.py runserver

You can now open up localhost/8000/cityfinderapp in your browser and see the front-end

Code description

All relevant project code is within cityfinder directory. /scripts contains code we used to clean or transform the data to put it into django's database. db_transform.py is legacy code, while import_data.py was the latest we used to update the database.

/cityfinderapp contains most of our work, including the Django application. algorithms.py and City.py are files with code we created to run a weighted-scores algorithm to rank cities according to user preference.

Data

We downloaded our data in csv format from

  • US Census
  • American Housing Survey for median rents data
  • WalkScore website
  • NOAA weather website
  • American Communities Survey
  • US Department of Justice, Uniform Crime Statistics
  • US Bureau of Labor Statistics for cost of living

These sites either have a user interface /portal or we were able to easily copy-paste the data into a csv.

We had to clean-up to be able to put in the database. We used short bash scripts as well as some manual work in Sublime text (regex find-replace) to delete columns, change cities and states which were clearly wrong (Chicago was listed in Indiana!). The more data cleaning we did, the more cities were able to match when we created the relationship to the City model. We wanted to get close to having as many cities as the walk score file has (108), although some files, such as crime data, did not provide information for many of our 108 cities.

TODO:

There are still many interesting things we'd like to do with this project that we were unable to finish by our deadline.

  • use sessions to save user input in the html so user can see it if she goes back pages
  • improvements to the algorithm
  • additional city data
  • more visualizations
  • having 2 users input preferences (partners)
  • cities near each other feature

CONTRIBUTORS