Topic Modeling Made Easy

Sanil Mhatre

Is your organization looking to unlock valuable insights for large amounts of text data for sources like social media posts, product reviews and survey results ? Have you already tried techniques like key words/phrases, sentiment analysis or word cloud, but are looking for deeper insights? Topic modeling will elevate your Natural Language processing/Text Analytics to the next level and help your organization discover valuable business insights.

This demo-filled session uses a real life use-case of Team Health Survey data to introduce attendees to the concepts, tools and techniques of topic modeling. Using Jupyter notebook, we will walk through the Python code demo for:
• Data loading, clean-up and pre-processing
• Tokenization, stop words, n-grams and lemmatization
• Iterate through model training and evaluation
• Visualize output and interpret human readable topics
• Business value and application of topic modeling