Topic-Over-Time-LDA

This repo make use of LDA models to discover how health issues over the world evolve over time. The discovery is done using tweets from 3 major health channels (BBC,CNN,CBC) from 2013 to 2015.

Concepts:

1. Data Cleaning
2. Data pre-processing
3. Extra pre-processing to cater to our dataset
4. Visualisation of topic-word distribution with pyLDAvis and wordcloud
5. Visualisation of topic over time with seaborn