LDA_topicmodelling

Inspiration

Reddit comments can get lengthy and often form their own topics. Users end up using search engines such as google to search for relevant topics. A lot of the activity in reddit chat rooms about stocks and cryptocurrency can be used to detect and correlate possible price fluctuations.

What it does

This project aims to find the best number of clusters for the points in the dataset and label each point to a topic. Unsupervised learning algorithm LDA for NLP topic modelling is used here.

Challenges we ran into

• Time and hardware computational constraint.
• Debugging and troubleshooting error messages
• Could not explore and compare with other unsupervised learning techniques

Accomplishments that we're proud of

• Able to learn more about NLP techniques
• Able to learn more about unsupervised learning
• Able to learn more about topic modelling
• Able to learn more about packages

What we learned

What's next for LDA_topicmodelling

Built With

• gensim
• jupyter
• kaggle
• nltk
• notebooks
• numpy
• pandas
• python
• spacy