Filter sales to 1 row per customer per year.

Intended to run before make_history to ensure that multi-year/lifetime sales are accounted for. The default (and intended purpose) is to pick the maximum "duration" value per customer-year. Optionally, it will also pick the minimum value of month (intended for mid-year dashboards) if first_month = TRUE.

rank_sale(sale, rank_var = "duration", grp_var = c("cust_id", "year"),
  first_month = FALSE)

Arguments

sale	data frame: Input sales data
rank_var	character: name of variable(s) to use for ranking
grp_var	character: name of variable(s) used for grouping
first_month	logical: If TRUE, also ensures the output contains the earliest month by grp_var (intended for mid-year dashboards)

See also

Salic Function Reference: salic

Other license history functions: history_check, make_history

Examples

library(dplyr)
data(lic, sale)

sale_unranked <- inner_join(lic, sale)
#> Joining, by = "lic_id"
sale_ranked <- rank_sale(sale_unranked)

# check sale ranking - highest duration will always be picked
left_join(
    count(sale_ranked, duration),
    distinct(sale_unranked, cust_id, year, duration) %>% count(duration),
    by = "duration",
    suffix = c(".ranked", ".unranked")
)
#> # A tibble: 3 x 3
#>   duration n.ranked n.unranked
#>      <int>    <int>      <int>
#> 1        1    76717      78923
#> 2        3     2706       2710
#> 3       99     1810       1810

# with earliest month included
sale_ranked <- rank_sale(sale_unranked, first_month = TRUE)
left_join(
    count(sale_ranked, month),
    distinct(sale_unranked, cust_id, year, month) %>% count(month),
    by = "month",
    suffix = c(".ranked", ".unranked")
)
#> # A tibble: 12 x 3
#>    month n.ranked n.unranked
#>    <int>    <int>      <int>
#>  1     1     1582       1582
#>  2     2     2074       2093
#>  3     3     5536       5620
#>  4     4     6987       7138
#>  5     5     7465       7646
#>  6     6     9965      10543
#>  7     7    15876      17270
#>  8     8    10571      12285
#>  9     9     9102      12071
#> 10    10     6637       8981
#> 11    11     3682       5699
#> 12    12     1756       2730