Twitter tweets are retrived using tweepy library and given as input for a model trained on twitter sentiment dataset by kaggle. The model used is of the following structure and accuracy.
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
embedding (Embedding) (None, 141, 13) 329303
_________________________________________________________________
global_average_pooling1d (Gl (None, 13) 0
_________________________________________________________________
dense (Dense) (None, 3) 42
=================================================================
Total params: 329,345
Trainable params: 329,345
Non-trainable params: 0
_________________________________________________________________
This is the link to my kaggle kernel where I have documented the training process in brief.
The model has been used to see the recent sentiment among the twitter population for a given movie or political affair or any hashtag for that matter. The following shows the gui for this usecase.
