Add geographical methods

[MikeJohnPage]

• Should this package be able to work with spatial data (e.g., sf)?
• Should geographical functions such as this be incorporated?

[matthewgthomas]

Aggregating indices to different levels of geography is a key part of making an index, so let's move those geographical aggregation functions into this package.

{compositr} doesn't necessarily need to work with spatial data though. The aggregation functions just need lookup tables (provided by {geographr}) and population counts (provided by {geographr}) but I can't think of any need for boundaries and other spatial data.

[MikeJohnPage]

Should calculate_extent() and calculate_proportion() be removed from geographr once they are added to compositr (I think so)?

[matthewgthomas]

Yes please. They're also separately implemented and used in IMD, so can be removed from there too once added to compositr.

[MikeJohnPage]

I've been reviewing/refactoring the existing methods, and am confused about what you currently have for calculate_proportion: https://github.com/humaniverse/geographr/blob/ed416a6735f7c4d8da0b866260369e84c27319a2/R/calculate.R#L99

@matthewgthomas

1. Why do you limit the calculation of proportions to only two possible values?
2. The function description say "Calculate proportion of small areas in the higher-level geography that are within the 10% most deprived areas in the nation." Yet, the function appears to just calculate the proportions of different groups within a higher-level geography. What is the goal of the function? What should it achieve?

[matthewgthomas]

The main purpose of the function is to aggregate data from a smaller unit (e.g. LSOA) into higher-level units (e.g. LA).

The original goal of the function was to calculate the proportion of X% most deprived/vulnerable/etc. areas within a higher-level geography - see https://github.com/humaniverse/IMD/blob/master/R/proportion.R

I've completely forgotten why I generalised this function beyond its original purpose - and I clearly forgot to change the function description to reflect that it was generalised.

[MikeJohnPage]

Which version do you want adding in compositr? Generalised?

Would a version which allows the user to set the percentage threshold be helpful (e.g., 10% most deprived, 20% most deprived, etc.)?

[matthewgthomas]

Let's use the generalised version.

Would a version which allows the user to set the percentage threshold be helpful (e.g., 10% most deprived, 20% most deprived, etc.)?

This would be brilliant, thanks.