Corporate Fraud, LDA, and Econometrics — DSSG talk, 2019

Corporate Fraud, LDA, and Econometrics — DSSG talk, 2019

These slides were presented at Data Science Singapore’s March 2019 meeting. They cover:

  1. What is misreporting?
  2. Implementation of Latent Dirichlet Allocation (LDA) for predicting misreporting
  3. Validating LDA for use on business documents
  4. Econometrics for windowed approaches to backtesting
  5. Dealing with sparse events for logistic regression

Slides are available in an interactive web-based format (reveal.js) and as a pdf. A video of this talk is also publicly available, hosted by DSSG on Youtube.

revealjs pdf

More details are available in my paper with W. Brooke Elliott and Nerissa Brown, available on SSRN:

pdfssrn