https://github.com/mikata-project/filternet
A PyTorch ensemble neural network model used for time series analysis.
https://github.com/mikata-project/filternet
Last synced: 11 months ago
JSON representation
A PyTorch ensemble neural network model used for time series analysis.
- Host: GitHub
- URL: https://github.com/mikata-project/filternet
- Owner: Mikata-Project
- License: mit
- Created: 2018-11-01T17:13:21.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2019-03-05T18:53:08.000Z (over 7 years ago)
- Last Synced: 2025-04-14T23:41:26.307Z (about 1 year ago)
- Language: Python
- Homepage: https://mikata.dev/
- Size: 15.6 KB
- Stars: 59
- Watchers: 65
- Forks: 24
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
## FilterNet
FilterNet is a ensemble neural network model used for time series analysis. It is comprised of a 1D convolutional neural
network and fast.ai's MixedInputModel. An example of the network implemented in PyTorch is located in filternet.py and
provides the model class along with its corresponding dataset class. Links to the abstract, slides, and video of our
presentation at PyData Los Angeles 2018 will be provided below.
This 1D convolutional neural network (CNN) was inspired by the traditional use of filters in discrete time signal
processing. While developed independently, it closely resembles the findings described in the
[WaveNet paper by Borovykh et al](https://arxiv.org/pdf/1703.04691.pdf). While the 1D CNN performed well on its own,
datasets can have a lot of context associated with them (hour of day, day of week, etc.) which the 1D CNN alone is
unable to handle. We utilized [fastai's MixedInputModel](https://github.com/fastai/fastai), which has been used
successfully for tabular data, to include learnings on the context portion of our datasets. The two neural networks are
combined using a final regression layer and were found to compliment each other. In testing, the resulting ensemble
model outperformed one of our current best production time series models ([TBATS](https://robjhyndman.com/hyndsight/forecasting-weekly-data/)).
Our hope is by open sourcing our approach it will help generate further ideas on how to improve time series modeling
using neural networks.
PyData Los Angeles 2018 presentation:
- [Abstract](https://pydata.org/la2018/schedule/presentation/14)
- [Slides](https://docs.google.com/presentation/d/e/2PACX-1vR6eea4L_Z_hyz24kgch3Lt5eEQ9PmmI2gUys_DcQrWY0EbG5CfOy4suqeLejXEql3x-nYT2NshrQRc/pub?start=false&loop=false&delayms=3000)
- [Video](https://www.youtube.com/watch?v=nMkqWxMjWzg)
Contributing authors:
- Jeff Roach (Data Scientist at System1)
- Nathan Janos (Chief Data Officer at System1)
For more information about System1, please visit: www.system1.com