Twitter_Sentiment_Analysis

- Keras
- Scikit Learn
- Pandas
- MatplotLib
- Beautiful Soup
- The dataset contains 1600000 rows with 6 coulmns
- Dropped unwanted columns,Preprocessed Sentences from the dataset
- Used transfer Learning for embedded matrix Glove 100d
- The model contain a mix of Conv1D,BiDirectional LSTM,Dense Layers followed by Dropout,BatchNormalization layers
- Achieved ~78% Training accuracy, ~80% Test accuracy(model ran for 5 epochs)
- Got F1 Score of 0.80
- converted unconevrted HTML with BS4
- Removed usernames,Urls,BOM's
- Removed #,Numbers
- Removed StopWords