Trying to predict average game rating. Use column ['score_phrase'] in sentemental layer, ['platform', 'genre', 'editors_choice', 'release_year'] - as categorical properties.
Build NN with more than one inputs, and more than one layers to merge.
- pandas
- tflearn
- tensorflow
- numpy
- sklearn
#Result
Achitecture:
Loss:
Loss(step 34k-58k):
Loss(step 70k-145k):
Loss(step 70k-200k):
- Training Step: 38334 | total loss: 1.06394
- Training Step: 144524 | total loss: 0.73067
- Training Step: 200068 | total loss: 0.52506