diff --git a/README.md b/README.md index 2ebd45a..732379e 100644 --- a/README.md +++ b/README.md @@ -3,9 +3,9 @@ Team project for Data Ethics (BA840) ### Data Source -[IRS Audit Rates by County from ProPublica](https://www.propublica.org/datastore/dataset/irs-audit-rates-by-county) -[Tax Return Data by County from IRS](https://www.irs.gov/statistics/soi-tax-stats-county-data) -[US Census Demographic Data from Kaggle] (https://www.kaggle.com/muonneutrino/us-census-demographic-data) +- [IRS Audit Rates by County from ProPublica](https://www.propublica.org/datastore/dataset/irs-audit-rates-by-county) +- [Tax Return Data by County from IRS](https://www.irs.gov/statistics/soi-tax-stats-county-data) +- [US Census Demographic Data from Kaggle](https://www.kaggle.com/muonneutrino/us-census-demographic-data) ### Project Objectives - Inspect the fairness of the IRS (Internal Revenue Service) audit algorithm. We would like to know: @@ -14,7 +14,7 @@ Team project for Data Ethics (BA840) - Is the algorithm fair in how it assigns audits? ### Project Summary --By exploratory data analysis, we examine the distribution and relationship between the audits and the demographic variables (race, gender, income level, etc). We found that race and socioeconomic status could be the proxy variables in the audit algorithm, even though the algorithm does not explicitly consider them. +- By exploratory data analysis, we examine the distribution and relationship between the audits and the demographic variables (race, gender, income level, etc). We found that race and socioeconomic status could be the proxy variables in the audit algorithm, even though the algorithm does not explicitly consider them. - In terms of consequentialism and deontology: - Consequentialism: The algorithm doesn't result in a good outcome on balance. Lower-income taxpayers face a higher chance of an audit. If the audit algorithm is unfair, it will result in certain groups of people will be treated unfairly in terms of time input in filing tax return documents.