shydefoo / topic-modelling-101

Topic modeling crash course.

Method: Latent Dirichlet Allocation Corpus used: 100 thousand articles from WikiCorpus Number of topics: 100

Create conda environment:

conda create --name lda101 python=3.6

Install requirements

pip install -r requirements.txt

Use lda_model_driver.ipynb test topic modelling

Things to change:

• repo_path repo_path = '' replace repo path with location repo was cloned
• Change url to a nytimes article and run cell to view topics scores and associated words.
• Note, scraper might be buggy. This is just a 101