View Code? Open in Web Editor
NEW
Sentiment Extraction on Social media texts related to Cryptocurrencies.
blockchain_project's Introduction
Sentiment Analysis using Naive Bayes, BERT and RoBERTa
- The dataset contains around 50k+ Annotated tweets.
- The labels are unbalanced so, we balance the data by picking equal number of each sentiment.
- 5k Positive tweets
- 5k Negative tweets
- 5k Neutral tweets
- Dataset cleaning
- Removing NaNs
- Balancing the dataset
- Removing unnecessary words/names of Cryptocurrencies
- Visualizing the data
- Modelling
- Observations and conclusion
blockchain_project's People
Contributors
Watchers