-
This project performs Sentiment Analysis of sample Twitter Data set.
-
Training data consists of a column called Sentiment which is a binary value.It is 1 if the corresponding tweet is positive and is Zero if its negative.
-
To build a Neural Network model,top 5000 tweets are considered and they are tokenized.
-
Above 5000 tweets are then divided into training (67%) and validation (33%) sets.
-
Above data is fit into a RNN model built with 5 epochs.
-
Then a LSTM model is built by replacing above RNN layer with LSTM layer.
-
Finally,a comparision of the results from LSTM and RNN models is done.