In this notebook I did sentiment analysis on tweeter dataset.
I designed 4 model for this purpose:
- Bi-directional LSTM
- Convolutional Nural Network
- Logistic Regression
- Baysian Classifier
For Deep Learning models I used Word Embedding Keras layer for feature extraction and for classic models I used CountVectorize and TF-IDF.
At the end I concluded that Bi-LSTM layer hast better performance for this dataset.