Example for using Tensorflow / Keras with a custom callback

Lately i switched from using XGBoost to using Tensorflow / Keras for my models in the tournament, but still wanted to have the metric output like in the era boosting example. Additionally, the callback saves the best model based on these metrics.

So i created a custom callback to show the metrics after each epoch while training. If you want to take a look, i shared it on Github:

If you have any comments or improvements please let me know.


What is your reasoning behind switching from XGBoost to a neural net?

Short answer: results :wink: