title Feature aggregators

When designing a diseasystore (see vignette("extending-diseasystore")), each feature must specify how that data should be summarised when joined with other data.

The automatic coupling and aggregation is essentially a three part process: First, data for the observable is joined with data required to form the stratifications. Next, the combined data is grouped at the level of stratification requested. Finally, the data is summarised in each group to form a single value for the group.

Below, we provide a set of simple function that perform this final summarisation step.

For semi-aggregated data, a typical key_join aggregator is the key_join_sum() which just adds the observations across the group

In some cases, such as if the observable is the "maximum temperature" the relevant aggregator would be the key_join_max() aggregator which takes the max across the group.

When working with individual level data, the observables often consists of a record per individual and to summarise the data we instead need the key_join_count() aggregator to get the size of the group.

Usage

key_join_sum(.data, feature)

key_join_max(.data, feature)

key_join_min(.data, feature)

key_join_count(.data, feature)

Arguments

.data

(any)
The data object to perform the operation on.

feature

(character)
Name of the feature to perform the aggregation over

Value

A dplyr::summarise to aggregate the features together using the given function (sum/max/min/count)

Examples

  # Primarily used within the framework but can be used individually:

  data <- dplyr::mutate(mtcars, key_name = rownames(mtcars), .before = dplyr::everything())

  key_join_sum(data, "mpg")    # sum(mtcars$mpg)
#>   value
#> 1 642.9
  key_join_max(data, "mpg")    # max(mtcars$mpg)
#>   value
#> 1  33.9
  key_join_min(data, "mpg")    # min(mtcars$mpg)
#>   value
#> 1  10.4
  key_join_count(data, "mpg")  # nrow(mtcars)
#>   value
#> 1    32