326 forward link

[AntonZogk]

Summary

get_link function for forward or backward imputation
Zerofy_values for ignoring values when calculating links based on a expression

Checklists

This pull request meets the following requirements:

• installable with all dependencies recorded
• runs without error
• follows PEP8 and project specific conventions
• appropriate use of comments, for example no descriptive comments
• functions documented using Numpy style docstings
• assumptions and decisions log considered and updated if appropriate
• unit tests have been updated to cover essential functionality for a reasonable range of inputs and conditions
• other forms of testing such as end-to-end and user-interface testing have been considered and updated as required
• tests suite passes (locally as a minimum)
• peer reviewed with review recorded

Both of these are not applicable since this pull request relates to intermediate functions for the impute method

• installable with all dependencies recorded is not applied here
• other forms of testing such as end-to-end and user-interface testing have been considered and updated as required

[AntonZogk]

• zerofy_values will update the input dataframe, let me know if you think it should return a new series or dataframe
• i included some basic scenarios in the unit tests because i think at the moment we don't have a clear structure and if i add csv files to tests will start getting messy, we can easily add more scenarios as needed

[Jday7879]

Great work Anton, I've left a few suggestions and comments. Happy to discuss further

[Jday7879]

It looks like the bulk of this dataframe is used a few times in the unit tests. I agree that having fewer files to load makes the repo cleaner, but having one csv which has the setup would remove the repeated setup. I also appreciate that the "question" column changes depending on the filter applied which makes it more difficult to use a csv

[AntonZogk]

I agree with you about this, another point is according to pytest docs

"Something to be aware of when grouping tests inside classes is that each test has a unique instance of the class. Having each test share the same class instance would be very detrimental to test isolation and would promote poor test practices"

Not sure about appropriate path to save it though, perhaps we should save them in a folder test_data ?

[Jday7879]

Other functions return dataframe, but I think the point about returning dataframe or series is a good one and should probably be discussed (i.e. how will all of the functions link together). Happy for this to be left as is and a ticket added to the backlog as other functions might need to be refactored

[AntonZogk]

Made some changes in the test approach, also please ignore the mask_values since needs further discussions

[Jday7879]

One minor point to consider, happy for it to be commit as is but might be worth to raise with another team member to get their input

[Jday7879]

Happy for code to be merged following minor changes. Nice work Anton!