Review and Checks

[HadleyBN]

Review old issues
These are issues created by Annabel that need to be reviewed to see if they are still relevant

• Null values for confidence statistic
• Reference data join columns
• Wilson and numerator < denominator? check error handling in R package..
• Wilson is the same calculation? get rid of tuples like R?
• Can R do DSRs for not all quantiles like the first test suggests?
• findxvalues unable to handle NA values within xvals supplied

Other Changes

• Decide if all (or any) of the confidence interval functions should be exported (currently they are all exported). Speak to Sam and Georgie regarding this issue. Personally, I think we should only export functions exported in PHEindicatorMethods for consistency.
• Decide if all (or any) of the funnels functions should be exported
• Based on above decision, create generic dobsons() function (currently only have dobsons_upper() and dobsons_lower() for consistency with other confidence interval functions. Will not be required unless all confidence interval functions remain exported.

Additional Review

• Check that df.copy() is being applied to all exported function to ensure that we create a copy of the dataframe being passed so that we don't change the users dataframe. It should be applied in all function that manipulate a dataframe in someway. Most of the functions will have it applied in validate data, but some may not use this function. Look out for functions where some operations are carried out before validations, such as in DSR.py.
• General pass to tidy up comments, there are still a couple of comments labelled as TODO. Assess these and move them to issues if they need to be actioned.