NPLAM

This is a Tensorflow implementation of the paper: Neural Partially Linear Additive Model.

Requirements

The model is implemented using Python3.7 with dependencies specified in requirements.txt.

Data Preparation

Simulation datasets

All simulation dataset are generated by createdata.py.

Real-world datasets

Download the California Housing dataset, the Super-conductivity dataset, the Beijing Air Quality dataset, and Boston Housing dataset from https://www.dcc.fc.up.pt/~ltorgo/Regression/cal_housing.html, https://archive.ics.uci.edu/ml/datasets/Superconductivty+Data, https://archive.ics.uci.edu/ml/datasets/Beijing+Multi-Site+Air-Quality+Data, and https://archive.ics.uci.edu/ml/machine-learning-databases/housing/, respectively.

Move them into the data folder.

Read dataset

All data can be loaded from "datasets". "read_data.py" can read the data and divide this dataset into training set, validation set and test set.

Model Structure

Layer and Model

• The layers of the NPLAM are shown in the layer.py.

• The model of the NPLAM are shown in the model.py.

Get the result

• result.py and run.py are the main files for running NPLAM.