unit8co/darts

[New model] DeepVAR

Open

#657 建立於 2021年12月2日

在 GitHub 查看
 (0 留言) (2 反應) (0 負責人)Python (6,832 star) (762 fork)batch import
good first issuenew model

描述

https://arxiv.org/abs/1910.03002

TBD if we can make it part of our current RNNModel (DeepAR) or if we need a new model.

貢獻者指南

技術棧
python
領域
machine learning
議題類型
feature
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
5
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
over 1 week
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
needs investigation
前置要求
Understanding of DeepAR modelFamiliarity with darts codebaseKnowledge of time series forecastingRead paper arXiv:1910.03002
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
15
研究方向
This issue proposes implementing the DeepVAR model from the paper (arXiv:1910.03002) for multivariate probabilistic time series forecasting. The first step is to study the paper and the existing DeepAR implementation in darts (likely in 'darts/models/forecasting/rnn model.py') to determine if DeepVAR can be integrated into the existing RNNModel class or requires a new model. Review the repository's contributing guidelines and existing model structure to plan the implementation. No prior discussion or linked PRs exist, so the contributor should initiate a discussion on the approach before coding.