Attached Jupyter Notebook is the part 5 of the Twitter Sentiment Analysis project I implemented as a capstone project for General Assembly's Data Science Immersive course.
Accompanying blog posts can be found from my Medium account: https://medium.com/@rickykim78
Below implementations can be found in the attached notebook.
experiment with
- varying vocabulary size
- n-grams
various models has been tested with the features extracted from Tf-Idf vectorizer
sentiment classifier using custom-defined positivity metric based on lexical approach