💡 Feature Request - From Wide to Long and vice-versa Longitudinal data formatting, inspired from TIDAL

[simonprovost]

🌟 Howdy Innovator! Got a groundbreaking idea for Scikit-longitudinal? We'd love to hear it! 🌟

Before you submit this feature request, consider the following:

• Have you checked if this feature already exists or has been requested before?
• Does this feature align with the goals and scope of Scikit-longitudinal?

If you do not know about the above, feel free to open a question based Github-issue.

🤔 What's the Problem?

Many longitudinal datasets are available in either long or wide formats. Scikit-longitudinal deals with wide-format longitudinal datasets, as indicated in the documentation (https://scikit-longitudinal.readthedocs.io/latest/tutorials/overview/). However, switching from one to another (Wide to Long, or vice versa) would be adding a lot of flexibility from a user viewpoint to any incoming datasets.

It is now impossible since the input data must already be in a wide format.

🎯 The Dream Solution

Investigate what TIDAL greatly did at https://tidal.shinyapps.io/tidalapp/ (Data preparation tab) by @kwongsiufung, @EileenYXu, and @AmeliaES and integrate a similar process in the LongitudinalDataset utility class: https://scikit-longitudinal.readthedocs.io/latest/API/data_preparation/longitudinal_dataset/

PS: Saving this before I forget about such a nice trick to add to Sklong.

Cheers

[simonprovost]

BONUS

• Let's overcome limitations like It is recommended your longitudinal dataset has at least 4 time points. This must be less than 400 MB.
• Let's use ETL libraries like Polar to intelligently cover a lot of NaN patterns. avoid column names must not have any spaces in them and missing data must be coded as "NA" kind of limitations.