Twitter-Sentiment-Analysis-Using-LSTM and Gensim

In this notebook, I have implemented Stacked LSTM with embedding to analyse 1.6Million tweets which is divided into three categories 1. Positive 2. Negative 3. Neutral, made model to predict class of new tweets with accuracy of 78 percent.

Final results after deployment

Performance

	precision	recall	f1-score	support
0	0.78	0.75	0.76	79800
1	0.76	0.79	0.77	80200
accuracy			0.77	160000
macro avg	0.77	0.77	0.77	160000
weighted avg	0.77	0.77	0.77	160000

Using Gensim

Model

Datasets

• Stanford's GloVe 100d word embeddings : https://www.kaggle.com/danielwillgeorge/glove6b100dtxt/tasks

Paper :

https://www.academia.edu/35947062/Twitter_Sentiment_Analysis_using_combined_LSTM_CNN_Models