Modeling categorical features using NN's
This project contains a small sample of a neural net modeled with Keras. The gist of the project is that categorical features are modelled by replacing each categorical feature with a mini net built of one dummy feature per level and on top one linear node. See the image for a visual hint:
This is a practical example of how NN can be made to shape data. The code uses the Keras model functional API to build the model.
The data files are not included, these are part of a Kaggle competition.
- pydot 1.0.2