Run decomposition across groups and decomp years

Applies the Shapley decomposition to each group × decomp-year in the data.

Usage

run_decomp(data, params)

Arguments

data

A data.table with one column per decomp variable (0/1 indicator), one column per summary variable, and optionally a decomp_year column and the group_vars columns. Each group must have exactly one row per on/off combination; row order does not matter.

params

A list with at least:

summary_vars

Character vector of summary variable column names

decomp_vars

Character vector of factor column names (0/1 indicators)

decomp_method

Character, "shapley" ("das_gupta" accepted as a deprecated alias)

group_vars

Optional character vector of grouping columns, such as a location identifier

Value

A data.table with one row per group × decomp-year × measure, and one column per factor giving its decomposition effect.

Examples

# Outcomes for every on/off combination of two factors, where the
# factors interact: y = 10 + 5a + 3b + 2ab
rt <- make_run_table(c("a", "b"))
rt[, y := 10 + 5 * a + 3 * b + 2 * a * b]
#>    run_id     a     b     y
#>     <int> <int> <int> <num>
#> 1:      1     0     0    10
#> 2:      2     1     0    15
#> 3:      3     0     1    13
#> 4:      4     1     1    20
rt[, decomp_year := 2000L]
#>    run_id     a     b     y decomp_year
#>     <int> <int> <int> <num>       <int>
#> 1:      1     0     0    10        2000
#> 2:      2     1     0    15        2000
#> 3:      3     0     1    13        2000
#> 4:      4     1     1    20        2000
run_decomp(rt, list(summary_vars = "y", decomp_vars = c("a", "b"),
                    decomp_method = "shapley"))
#>    decomp_year measure     a     b
#>          <int>  <char> <num> <num>
#> 1:        2000       y     6     4