Ethnicolr 2023 Paper Replication Materials

Data

• FL 2022 Voter Registration Data

Scripts

1. Data Preprocessing

   • Full Name
   • Last Name

2. Random Forest

   • Full Name Rapids Unigram
   • Full Name (with bigram)
   • Last Name

3. Gradient Boosting

   • Full Name
   • Last Name

4. LSTM

   • Full Name
   • Last Name

5. Transformer

   • Full Name
   • Last Name

6. KNN Cosine Distance

   • Full Name
   • Last Name
   • See also Github Repository for stuff like LSH Minhash implementation.

7. Multi Output Regression

   • Last Name (MLP)
   • Last Name (LSTM)

8. LSTM with Synthetic Data Last Name

   • Synthetic Data. Output here
   • Model

9. Validation

   • North Carolina
   • Census
   • Rosenman et al.

10. Applications

    • Diversity of Campaign Donations
      • DIME from https://data.stanford.edu/dime
      • Analyses
    • Diversity in News
      • https://github.com/notnews/rainbow

11. Interpretation

    • RF

12. Create Tables

13. Uncertainty

    • https://github.com/appeler/uncertain_names

Models

• Models
  • Excludes large RF models. We don't see the point of including models that are not very computationally expensive but have very large model files.

Tables

• Tables

MS

• Manuscript

History

The 2018 version of the paper and pres. is here.

Authors

Rajashekar Chintalapati, Suriyan Laohaprapanon, and Gaurav Sood