bayeshack2016/financial-marketplace-referee

Name: financial-marketplace-referee

Owner: Bayes Hack 2016

Description: #commerce #DOC Financial Marketplace Referee

Forked from: bayeshack16-superawesometeam/bayeshack16

Created: 2016-04-24 16:05:25.0

Updated: 2016-10-18 03:50:15.0

Pushed: 2016-04-26 17:02:39.0

Homepage:

Size: 62049

Language: Jupyter Notebook

GitHub Committers

UserMost Recent Commit# Commits

Other Committers

UserEmailMost Recent Commit# Commits

README

Financial Marketplace Referee (FMR)

Context

The Consumer Financial Protection Bureau (CFPB) publishes a public consumer complaint database consisting of complaints relating to financial services products and companies. Unscrupulous companies sometimes take advantage of consumers through predatory or discriminatory product offerings and targeting, leading to the consumers lodging complaints against these products and companies to the CFPB.

By enriching the consumer complaint dataset with population demographic statistics sourced from the American Community Survey, our research shows that there exist notable relationships between companies that have received complaints from those geographical regions with the largest minority populations (on the 5-digit zipcode granulatity) and the companies that are eventually charged with discriminatory product offerings.

Most notably, we show that our analysis can help identify companies that possibly offer predatorial, fraudulent, and racially discriminatory financial products. This can help influence policy and investigation decisions by financial regulatory bodies, and help consumers get better access to data about the suspiciousness of a financial vendor.

Description

Financial Marketplace Referee is an application toolset that performs analysis on the latest CFPB consumer complaint database and ASR zipcode demographic statistics to inform the user of companies that are suspected to be involved in financial foul play.

There are two analysis modules that can be executed:

  1. Complaint Volume
    • This module analyzes the data without reference to demographics and flags company suspiciousness based on the number of complaints lodged against it.
  2. Racial Discrimination
    • This module attempts to find patterns of complaints that could indicate a company is discriminating against minority groups. Joining against the ACS database, this analysis module finds companies with abnormally high numbers of complaints from geographic regions with the largest minority populations. This module can both find potentially risky companies for a given product line, and can score a single company on all product lines which they offer.

Each of these modules allow the user to ask two questions:

  1. Is company X bad?
    • This suggests to the user if the company is suspected to be offering “bad” products across a range of product lines.
  2. For a particular product line, (e.g. Mortgage, Student Loan) which companies should I avoid?
    • This presents the user a list of companies with the anomalous complaints that could indicate foul play.
Installation

First, clone the repository locally:

t clone https://github.com/bayeshack16-superawesometeam/bayeshack16.git

Next, install fmr:

thon setup.py install

Finally, enjoy!

Software Dependencies
Dataset Dependencies
Usage
Examples:

To find mortgage companies potentially discriminating against clients:

r \
--complaintsfile ./Consumer_Complaints.csv \
--demographicsfile ./zipcodes/output/zz_all_us_race.csv \
--racial \
--badcompanies \
--productline 'Mortgage' \
--pretty

Sample output:

al Analysis:

"Common Wealth Mortgage Services",
"First American Mitigators, PLLC.",
"Oceanside Mortgage Company",
...,
"Old National Bank"

To find out what JPMorgan Chase & Co. financial product lines are suspicious:

r \
--complaintsfile ./Consumer_Complaints.csv \
--volume \
--isbad \
--company 'JPMorgan Chase & Co.' \
--pretty

Sample output: ('true' means suspicious)

me Analysis:

"Debt collection": true,
"Prepaid card": false,
"Mortgage": true,
"Credit reporting": false,
"Student loan": false,
"Money transfers": false,
"Bank account or service": true,
"Credit card": true,
"Payday loan": false,
"Consumer Loan": true,
"Other financial service": true

For all options:

r -h

The following values are valid inputs for the “–productline” flag


'Student loan',
'Credit reporting',
'Credit card',
'Debt collection',
'Payday loan',
'Consumer Loan',
'Bank account or service',
'Mortgage',
'Money transfers',
'Prepaid card',
'Other financial service'
People
This work is supported by the National Institutes of Health's National Center for Advancing Translational Sciences, Grant Number U24TR002306. This work is solely the responsibility of the creators and does not necessarily represent the official views of the National Institutes of Health.