select_neighbours: Select Subset of Rows Closest to a Specified Observation
In pbiecek/ceterisParibus2: Ceteris Paribus Profiles

Description Usage Arguments Details Value Examples

This function selects subset of rows from data set. This is useful if data is large and we need just a sample to calculate profiles.

1 2	select_neighbours(data, observation, variables = NULL, distance = gower::gower_dist, n = 20, frac = NULL)

`data`	set of observations
`observation`	single observation
`variables`	names of variables that shall be used for calculation of distance. By default these are all variables present in 'data' and 'observation'
`distance`	the distance function, by default the 'gower_dist' function.
`n`	number of neighbours to select
`frac`	if 'n' is not specified (NULL), then will be calculated as 'frac' * number of rows in 'data'. Either 'n' or 'frac' need to be specified.

Note that select_neighbours function is S3 generic. If you want to work on non standard data sources (like H2O ddf, external databases) you should overload it.

a data frame with selected rows

library("DALEX2")

new_apartment <- apartments[1, 2:6]
small_apartments <- select_neighbours(apartments_test, new_apartment, n = 10)
new_apartment
small_apartments