Add Cosine, ACF and PACF transformers

[afzal442]

Reference Issues/PRs

Fixes #483

What does this implement/fix? Explain your changes.

Added two transformations.

Does your contribution introduce a new dependency? If yes, which one?

No

PR checklist

• I've added myself to the list of contributors.
• Optionally, I've updated sktime's CODEOWNERS to receive notifications about future changes to these files.
• I've added unit tests and made sure they pass locally.
• I've added the transformers series to the file.
• I've updated the existing file.

[mloning]

Hi @Afzal-Ind - thanks for submitting the PR! I had a quick look, the overall structure looks good, I left two comments in the code.

Next thing to do is to write unit tests for the transformers you added. For an example, check out the unit tests for the boxcox transformer

[mloning]

There were some merge conflicts, but I resolved them with the latest commit. To avoid them, you need to make sure to sync your fork before you check out a new branch,

[afzal442]

There were some merge conflicts, but I resolved them with the latest commit.

Hi @mloning. Thanks a lot... I will take care of this.

[mloning]

Hi @Afzal-Ind looks much better now - next step is to write some simple unit tests as described above!

[mloning]

Hi @Afzal-Ind - okay renaming solved one issue. Now the issue seems to be related to how multivariate data is handled. SeriesToSeriesTransformer can take a univariate series (pd.Series) or multivariate series (pd.DataFrame with rows representing time points and columns variables). We need to either (i) enforce univariate data and raise an error if multivariate data is passed by setting check_series(..., enforce_univariate=True) or (ii) make sure the transformers correctly handle multivariate data.

I'd start by setting check_series(..., enforce_univariate=True) to get this to run first, we can then add support for multivariate once everything is working.

[afzal442]

I'd start by setting check_series(..., enforce_univariate=True) to get this to run first, we can then add support for multivariate once everything is working.

Fine now. I can start trying this way. Thanks.

[mloning]

Hi @Afzal-Ind if you have time to work on this, we could try to include it in the next release which I'm currently working on. For now, we can just remove the inverse_transform.

[afzal442]

Fine @mloning . I am looking forward to working with u on 'inverse_transformer' in next release. Currently dealing with AutoAR. Thanks.

[mloning]

@Afzal-Ind I want to push the next release in the next few days, so if you have time to finish it, that would be great. Otherwise we'll include in the next one after that. Of course let me know if you get stuck and I can help!

[afzal442]

Of course let me know if you get stuck and I can help!

I was kind of busy for two days. I would start my work on that from tomorrow and let u update.

[mloning]

Hi @Afzal-Ind, I've taken a look and made some changes - hopefully all tests pass now. Take a look at my commit to understand what I changed - happy to discuss this if you have any questions!