Nice stuff – i used similar architecture but with 2 dropouts and more epochs (which took forever), also used Glove Embedding which didn’t help much.
If you are interested: https://www.datasciencesociety.net/fighting-propaganda-task1/

Of course you are more than welcome 🙂 At the end if the day we are here to collaborate and learn! (At least i am 🙂

Thank you

Hi team, thanks for this article. It would be great to copy paste your code on a Jupyter notebook and upload it as a media file (Jupyter comes for free with Anaconda Distribution so you can go ahead and install this).
I was curious about the clustering – based on what do you define to come up with 196 clusters? Furthermore how would you estimate the temperature, pressure, etc. for these clusters, when you reach to the prediction stage, are you going to take some sort of an average?

Thanks for sharing, there are some interesting concepts. For the distance algorithm you might want to check out the vincenty distance:
https://stackoverflow.com/questions/38248046/is-the-haversine-formula-or-the-vincentys-formula-better-for-calculating-distan