https://github.com/pyro-ppl/Pyro-M5-Starter-Kit

Learn Pyro through the M5 forecasting competition
https://github.com/pyro-ppl/Pyro-M5-Starter-Kit

Last synced: 18 days ago
JSON representation

Learn Pyro through the M5 forecasting competition

• Host: GitHub
• URL: https://github.com/pyro-ppl/Pyro-M5-Starter-Kit
• Owner: pyro-ppl
• License: apache-2.0
• Created: 2020-03-10T04:28:18.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-04-07T04:38:17.000Z (over 4 years ago)
• Last Synced: 2024-07-31T21:55:18.057Z (3 months ago)
• Language: Python
• Homepage:
• Size: 6.49 MB
• Stars: 83
• Watchers: 7
• Forks: 10
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Pyro M5 Starter Kit

This repo aims to help you learn the [Pyro](https://pyro.ai) probabilistic programming language through the [M5 forecasting competition](https://www.kaggle.com/c/m5-forecasting-uncertainty/overview).

## Documentation resources

- Pyro [reference docs](https://docs.pyro.ai)

  - [Installation](https://docs.pyro.ai/en/stable/installation.html)

  - [Forecasting](https://docs.pyro.ai/en/stable/contrib.forecast.html)

  - [Distributions](https://docs.pyro.ai/en/stable/distributions.html)

  - [Reparameterizers](https://docs.pyro.ai/en/stable/infer.reparam.html)

  - [Time series](https://docs.pyro.ai/en/stable/contrib.timeseries.html)

- Pyro [tutorials](https://pyro.ai/examples)

- Pyro [forum](https://forum.pyro.ai)  <---- ask questions here

- PyTorch [reference docs](https://pytorch.org/docs/stable/index.html)

- M5 competition [overview](https://www.kaggle.com/c/m5-forecasting-uncertainty/overview)

- M5 exploratory data analysis [notebook](https://www.kaggle.com/headsortails/back-to-predict-the-future-interactive-m5-eda) (thanks to Martin Henze)

I'd recommend starting with the following sequence of Pyro tutorials. These build up to [Forecasting III's Model2](https://pyro.ai/examples/forecasting_iii.html#Deeper-hierarchical-models) which is very close to the M5 forecasting problem.

1. [Pyro models](https://pyro.ai/examples/intro_part_i.html) *recommended*

2. [Pyro inference](https://pyro.ai/examples/intro_part_ii.html)

3. [Stochastic variational inference I](https://pyro.ai/examples/svi_part_i.html)

4. [Stochastic variational inference II](https://pyro.ai/examples/svi_part_ii.html) *recommended*

5. [Stochastic variational inference III](https://pyro.ai/examples/svi_part_iii.html)

6. [Tensor Shapes](https://pyro.ai/examples/tensor_shapes.html) *recommended*

7. [Forecasting I: univariate, heavy tailed](https://pyro.ai/examples/forecasting_i.html) *recommended*

8. [Forecasting II: state space models](https://pyro.ai/examples/forecasting_i.html)

9. [Forecasting III: hierarchical models](https://pyro.ai/examples/forecasting_i.html) *recommended*

## Contributions welcome!

We believe in friendly competition that builds on collaboration.