/Evaluating-LDA

Repo for my talk at the PyData Berlin 2017 conference

Primary LanguageJupyter NotebookMIT LicenseMIT

pydataberlin-2017

Repo for my talk at the PyData Berlin 2017 conference.

Talk is available on YouTube https://www.youtube.com/watch?v=UkmIljRIG_M

The talk covers evaluation measures for topic models including

  • visualisation tools
    • termite
    • pyldavis
  • metrics
    • perplexity
    • coherence (pipeline)
  • human judgement based methods
    • word intrusion
    • topic intrusion

Refer to the notebook for model training, creating the visualisation and running the evaluation metrics.