PyTorch Tabular – A Framework for Deep Learning for Tabular Data

It is common knowledge that Gradient Boosting models, more often than not, kick the asses of every other machine learning models when it comes to Tabular Data. I have written extensively about Gradient Boosting, the theory behind and covered the different implementations like XGBoost, LightGBM, CatBoost, NGBoost etc. in detail. The unreasonable effectiveness of Deep Learning that was displayed in many other modalities – like text and image- haven not been demonstrated in tabular data. But lately, the deep learning revolution have shifted a little bit of focus to the tabular world and as a result, we are seeing new architectures and models which was designed specifically for tabular data modality. And many of them are coming up as an equivalent or even slightly better than well-tuned Gradient Boosting models.