Congressional Money Trail

Learn more about the developer on LinkedIn.

Check it out live at congressional-money-trail.herokuapp.com

Campaign contributions to our federal officials are publicly disclosed but rarely easy to access. Utilizing the past decade of campaign contribution data, Congressional Money Trail seeks to help the public better understand from where their federal congressional representatives receive their campaign money. Just select a state or find a Member of Congress by address and a visual representation of how they finance their campaigns is revealed.

Features

Current

• Campaign Contribution data (pipe and comma separated format) from Opensecrets.org imported into PostgreSQL

• SQLAlchemy and raw SQL queries database for total campaign contributions to a user-selected Member of Congress.

• Flask-Python server sends JSON object to DOM.

• Result rendered as a tree-diagram graph (d3.js)

• Flask app renders HTML, Jinja2 and handles AJAX requests to the database

• Option to find a Member of Congress by state

• Option to search for Congressional delegation by address (Google GeocodeAPI and Sunlight Foundation CongressAPI)

• Nodes are sized based on proportion of contributions from identified category as compared to whole

• Summary of campaign contributions from particular category or entity as mouseover tooltips

Future

• Search Members of Congress by name
• Color lines/nodes according to industry of contributor
• Capture image to share on social media
• Extend data to include contributions to top PACs

Here is an example of a full visualization.

This is a screencast demo of the application - click to play on YouTube:

Examples of other stages of the app at the end of the document.

##Technology

Python, Flask, SQLite PostgreSQL, SQLAlchemy ORM, D3.js, jQuery, JS, AJAX, HTML, CSS, Sunlight Foundation CongressAPI, Google Geocode API

Dependencies are listed in requirements.txt.

To run the seed file and use the find congressional delegation by address function, you need both a GoogleMaps API key and Sunlight Foundation API key (both free).

##Backend and Database

Python runs the backend of Congressional Money Trail. Given the amount of data and speed necessary to deliver information to the DOM, I originally made my queries with both SQLAlchemy and raw SQL with indices added to the database to further increase querying speed. In addition I had much of the information and calculations needed to form the JSON object for the D3 visualization are pushed to Python rather than multiple queries.

Update (March, 2016): After noticing the data was being delivered slowly to the DOM in production, I have since created a new JSON column in the Legislator table that holds the necessary information. The JSON object is now created and loaded when you seed the database.

This means I will need to take that into account when I write an update function to update numbers when new data is available.

##Frontend

The front-end is composed using HTML, custom CSS combined with Bootstrap. Interactivity with the components are programmed with D3.js, jQuery, and AJAX.

D3.js was chosen for the visualization technology because of its prevalence in the data visualization realm and strong documentation.

##File Guide

• model.py Creates the tables for the database
• seed.py Seeds the database from source files
• server.py Controls the flask app
• TrailMapTree.js Handles the D3.js visualization rendering
• get_members.js Handles retrieving data on Congressional Delegations from database either via select-by-state option or select-by-address option.

##Installation

$ git clone [git-repo-url]
$ virtualenv env
$ pip install -r requirements.txt

Download datasets and save to a src subdirectory in project directory:

• Contribution data must be downloaded from [OpenSecrets.org] (http://www.opensecrets.org/myos/bulk.php) (free account required).

• Legislator information from [Sunlight Foundation] (https://sunlightlabs.github.io/congress/#legislator-spreadsheet)

• Make sure you have PostgreSQL installed and create a database named contributions

At the command line:

$ python -i model.py
$ db.create_all()

...then run

$ python seed.py

This will likely take awhile, depending on how many years of data you decide to include (see note on source files below).

To start server:

$ python server.py

###A note on campaign finance source files: Right now, the seed file is looking for the file handles for legislators.csv and CRP_Categories.txt (industry codes) in './src/'

For the contribution info - the seed file is looking for:

• Individual information in './src/individuals/' indivs04.txt through indivs14.txt

• PAC contribution information in './src/pac_to_cand/' pacs04.txt through pac14.txt

• PAC information in 'pac_info/' cmtes04.txt to cmtes14.txt

This may change to utilizing a config file in future versions.

##Additional Screenshots Homepage

Example of tooltip and view of individual donor branch

Search by address results

Location for user on information used for the site

Additional Full Visualization