unit8co/darts

Allow historical forecast to work on time series with nan

Open

#1,278 opened on Oct 11, 2022

View on GitHub
 (0 comments) (3 reactions) (0 assignees)Python (6,832 stars) (762 forks)batch import
good first issueimprovement

Description

We often want to see the historical forecast or back test on a time series having nan (we we can't really interpolate, because the missing part is just too long)

Consider returning nans or nothing for points which are not predictable (where some inputs are nan).

Might be also useful to create a TimeSeries property which indicates if there are some nan in the time series.

Contributor guide

Tech stack
pythonpandasnumpy
Domain
datamachine learning
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
basic knowledge of time seriesfamiliarity with darts libraryunderstanding of NaN handling
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
50
Research direction
Investigate how the HistoricalForecasting class currently processes time series data, specifically looking at the `predict` method and how it handles NaN values. Check if there is existing logic to propagate NaN or skip predictions for NaN inputs. Consider adding a property or option to allow NaN in historical forecasts, similar to how backtesting handles missing values. Look at related issues or PRs in the repository for prior discussions.
Allow historical forecast to work on time series with nan · unit8co/darts#1278 | Good First Issue