Inspiration

Sentiment Analysis Is becoming a hot topic nowadays and as usual, DeepLearning will Provide an end to end solution so I took a problem statement about Twitter Sentiment analysis and developed an end to end Deep Learning Model

What it does

Categorize tweets into positive or negative by using a Deep learning Model that learn from the training data

How I built it

Cleaned dataset with regular expressions and removed Stop words A custom architectured Neural Network that contains a mixture of 1D Convolution, LSTM and Bidirectional LSTM with regularization from batch regularization and Dropout layers.

Challenges I ran into

Cleaning DataSet

Accomplishments that I am proud of

Perfectly running Model with 80% test accuracy that trained only on 5 epochs

What we learned

Using Regular Expressions in an efficient way

What's next for Leverage Sentiment Analysis

Using Transformers

Built With

Share this project:

Updates