Nixtla / Nixtla/utilsforecast

Add ensemble method

Open
#220 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement feature
Dominant language
Python
Stars
109
Forks
26
Avg merge
3d 4h
Merged PRs (30d)
6

Description

Description

Current end points for forecasting (statsforecast, mlforecast, neuralforecaster) only allow usage of individual models. I am proposing the option for ensembling multiple models into one model output (as offered for example by AutoGluon).

I think, what would be an good starting point would be a simple average and the implementation of the greedy ensemble (local and global).

Open questions for me at this point are:

  • How to deal with prediction intervals (e.g.: conformal predictions) when combining different models?
  • Would there be a way to ensemble models from different repos like statsforecast and mlforecast?
Use case

No response

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the existing forecasting endpoints for statsforecast, mlforecast, and neuralforecaster to understand how individual models produce outputs. Define the scope for simple averaging and greedy local/global ensembles, including how prediction intervals and models from different repositories should be handled; completion requires those design questions to be resolved.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
data, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.