Feature Request: Add prewhitening routines

[adamkemberling]

Hi, thank you for your work implementing the Rodionov approach into a user-friendly R package. Are there any plans to incorporate the various prewhitening steps (OLS, MPK, IP4) that are described in Rodionov's 2006 paper? I'm not able to open/see their implementation in the workbook, but if I knew what was going on I may be able to code it up. If I can help I'm happy to contribute (since I likely will be coding it up anyways), I just need some confirmation that what I code up is true to the methodology.

Thanks,
Adam

[alexhroom]

hi Adam, thank you for your request! I'd left out the prewhitening because I assumed that some other package already existed to do it better. this is almost certainly the case with OLS but now I'm looking I can see that there aren't any existing R implementations of MPK and IP4 (apart from the ones in the long-abandoned rstars...).

Ideally I'd like to add any new features in Rust for speed (since the infrastructure is set up for the package anyway) but if you've coded up an implementation in R anyway then feel free to make a pull request, and I can rewrite it in Rust later on.

[adamkemberling]

Hi Alex, apologies for just asking for features and not giving much more context. I am working on a project and was selfishly looking for a faithful implementation of all of Rodionov's methods in R because it is a popular method in the literature and if that package existed already it would save time spent digging into the nitty-gritty of the implementation or the equations.

Before I came across the Stirnimann repo I had started to code up what I understood from the papers, but I hit some similar stumbling blocks and general uncertainty on whether it was a 1:1 match with the excel macro.

For the immediate time being I am using Luca Stirnimann's repo as a faithful reproduction of Sergei Rodionovs methods, but I'm adapting them slightly for my needs (ease of use changes like returning the data instead of saving files) but largely trying to keep them as-is to not get lost in the weeds. I don't know much about what needs to be done to implement it in rust, but I definitely don't want to step on your toes at all with your package development plans. If any of that seems consistent with what you had in mind, happy to help.