Time Series MLFlow logging during `setup`

[ngupta23]

Additionally, all the plots that doesn't require estimator can also be lagged at a setup stage such as ACF, PACF, ts, etc.

This is a good feature to have. We will take this up later (after model MLFlow logging is implemented).

Originally posted by @ngupta23 in https://github.com/pycaret/pycaret/issues/1782#issuecomment-955689962

[github-actions[bot]]

Stale issue message: This issue will be automatically closed by GitHub Actions in 1 week if there is no further activity.

[github-actions[bot]]

Stale issue message: This issue will be automatically closed by GitHub Actions in 1 week if there is no further activity.

[ngupta23]

We want to log the following plots during setup

• [ ] "diagnostics" (covers ACF, PACF, Periodogram, QQ-plot)
• [ ] "decomp" (covers classical decomposition)
• [ ] "diff" (covers difference plots)

[ngupta23]

This can mimic the way _mlflow_log_model is implemented

https://github.com/pycaret/pycaret/blob/b3fcf143f8d0dfae008fe8b69aef82074bec8215/pycaret/internal/pycaret_experiment/tabular_experiment.py#L168

So, maybe create another method called _mlflow_log_setup (called at the end of setup) and internally, you can call the respective plot_model calls (similar to what is done inside _mlflow_log_model

https://github.com/pycaret/pycaret/blob/b3fcf143f8d0dfae008fe8b69aef82074bec8215/pycaret/internal/pycaret_experiment/tabular_experiment.py#L293

[ryanxjhan]

Hi @ngupta23, can you assign this issue to me? Thanks!

[ryankarlos]

@ngupta23 Is this issue being worked on or can I have a go ? My first contribution to pycaret

[ngupta23]

Thanks @ryankarlos. I have assigned it to you. This is not being worked on right now by anyone.