Topic Modelling of NCAA College Sport News Coverage
Published:
Timeline: October 2025
Project Description: A pipeline that turns 173,999 raw newspaper articles into a clean, deduplicated corpus of about 150,000 articles about NCAA college sport from 2002 to 2025. It trains Gensim LDA topic models with 10 to 50 topics, scores them with c_v coherence, finds the main topic of each article, and tracks how coherent and how distinct each topic is from year to year. Built with Python, NLTK, Gensim and pyLDAvis on Google Colab.
Project Github Link: ncaa_analysis