Blogpost on this project: https://medium.com/@lukepersola/are-government-contractors-that-use-lobbyists-better-at-avoiding-penalties-215ec8e2c2f5
I worked on this project at Metis.
data/
contractor_misconduct.csv
- all data from The Project On Government Oversight's Federal Contractor Misconduct Databaseclean/
- output from notebooks- *
indicators
* - with indicator dimensions - the rest - compact, for analysis
- *
notebooks/
wrangle/
misconduct_clean.ipynb
- extract from CSV source and cleanadd_indicators.ipynb
- add indicators
analyze/
class_correlations.ipynb
- visualize correlations between categorical features and targetTSNE.ipynb
predict/
predict_disposition_type.ipynb
- survey of model performancepredict_fine.ipynb
- (work in progress) included)
Run:
misconduct_clean.ipynb
add_indicators.ipynb
disclosures_extract_clean.ipynb
add_lobbyist_features_to_misconduct.ipynb
- what you want from
analyze/
orpredict/