aggregate_key() produces aggregations of single leaf nodes

[mitchelloharawild]

This is not necessarily a bug, just something to be aware of. An option to produce unbalanced hierarchies for redundant aggregations may be useful.

library(fabletools)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
tsibble::tourism %>% 
  aggregate_key(Purpose * (State / Region), Trips = sum(Trips)) %>% 
  filter(Purpose == "Business", State == "ACT")
#> # A tsibble: 160 x 5 [1Q]
#> # Key:       Purpose, State, Region [2]
#>    Quarter Purpose  State Region       Trips
#>      <qtr> <chr>    <chr> <chr>        <dbl>
#>  1 1998 Q1 Business ACT   <aggregated> 150. 
#>  2 1998 Q2 Business ACT   <aggregated>  99.9
#>  3 1998 Q3 Business ACT   <aggregated> 130. 
#>  4 1998 Q4 Business ACT   <aggregated> 102. 
#>  5 1999 Q1 Business ACT   <aggregated>  95.5
#>  6 1999 Q2 Business ACT   <aggregated> 229. 
#>  7 1999 Q3 Business ACT   <aggregated> 109. 
#>  8 1999 Q4 Business ACT   <aggregated> 159. 
#>  9 2000 Q1 Business ACT   <aggregated> 105. 
#> 10 2000 Q2 Business ACT   <aggregated> 202. 
#> # … with 150 more rows

^{Created on 2020-06-19 by the reprex package (v0.3.0)}

[mitchelloharawild]

This may need an option, such as drop or trim which removes structurally redundant nodes in the hierarchy.

[mitchelloharawild]

MRE:

library(fabletools)
library(tsibble)
tibble::tribble(
  ~ grp_a, ~ grp_b, ~ time, ~ y,
  "a", NA, 1, 1,
  "b", "x", 1, 2,
  "b", "y", 1, 3
) %>% 
  as_tsibble(c(grp_a, grp_b), index = time) %>% 
  aggregate_key(grp_a/grp_b, y = sum(y))
#> # A tsibble: 6 x 4 [?]
#> # Key:       grp_a, grp_b [6]
#>    time grp_a        grp_b            y
#>   <dbl> <chr>        <chr>        <dbl>
#> 1     1 <aggregated> <aggregated>     6
#> 2     1 a            <aggregated>     1
#> 3     1 b            <aggregated>     5
#> 4     1 a            NA               1
#> 5     1 b            x                2
#> 6     1 b            y                3

^{Created on 2020-08-11 by the reprex package (v0.3.0)}