I’m currently working on a project where of course the question of neural networks comes up. We are working on a fairly simple regression problem that has some noise in the data. My current approach is testing a few different networks including Random Forest and XGBoost.
I did run some test with TensorFlow but I didn’t see much better results to justify the increased training time.
However I figured I should ask people a bit more seasoned than myself, why should I use TensorFlow for a simple regression algorithm on structured data?