https://github.com/harry24k/mida-pytorch
PyTorch implementation of "MIDA: Multiple Imputation using Denoising Autoencoders"
https://github.com/harry24k/mida-pytorch
autoencoder deep-learning imputation pytorch
Last synced: 6 months ago
JSON representation
PyTorch implementation of "MIDA: Multiple Imputation using Denoising Autoencoders"
- Host: GitHub
- URL: https://github.com/harry24k/mida-pytorch
- Owner: Harry24k
- License: mit
- Created: 2019-03-04T05:46:31.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2019-03-05T14:34:53.000Z (over 6 years ago)
- Last Synced: 2025-03-24T15:42:00.665Z (7 months ago)
- Topics: autoencoder, deep-learning, imputation, pytorch
- Language: Jupyter Notebook
- Homepage:
- Size: 39.1 KB
- Stars: 28
- Watchers: 1
- Forks: 8
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# MIDA-pytorch
**A pytorch implementation of "[MIDA: Multiple Imputation using Denoising Autoencoders](https://arxiv.org/abs/1705.02737)"**
## Summary
1. Doing imputation with Overcomplete AutoEncoder for missing data
2. Using complete data for training
3. Dropout is used to generate artificial missings in the training session
4. Experimenting with two missing methods(MCAR/MNAR)
5. Simple but good
## Requirements
* python==3.6
* numpy==1.14.2
* pandas==0.22.0
* scikit-learn==0.19.1
* pytorch==1.0.0
## Data
In the paper, 15 publicly available datasets used.
In this code, only 'Boston Housing' data is used among 15.
http://math.furman.edu/~dcs/courses/math47/R/library/mlbench/html/BostonHousing.html