Additionally, we decided to try reducing model noise from the previous models by cutting out certain features that were not well suited for neural networks and that were low scoring in terms of feature importance for the previous two models. The results were significantly better compared to the XGBoost model. Unlike the predictions of XGBoost, […]