Reproducing sentiment-free results

[HanyHossny]

Hi Ankitesh,

I am new to deep learning, and I think I missed something in reproducing the results of this code. I would appreciate if anybody can help.

I tried to run the code once with and without the sentiment data and I got the same mean absolute error:4.5762

I used the sentiment data from the file "data_senti_word.json" and I used the stock historical data only without sentiments by from the file "data.json". I changed the data dimension variable in "stockpred.py" to match the number of features in X.

I wonder, did I miss something to reproduce the results where sentiment-based scores are better than historical-data-based predictions.

Please advise,
Regards
Ahmad