Binning by value is the only original binning method implemented in this package. It is inspired by the case in marketing when accounts need to be binned by their sales. For example, creating 10 bins, where each bin represent 10% of all market sales. The first bin contains the highest sales accounts, thus has the small total number of accounts, whereas the last bin contains the smallest sales accounts, thus requiring the most number of accounts per bin to reach 10% of the market sales.
tibble::tibble(SALES = as.integer(rnorm(1000L, mean = 10000L, sd = 3000))) -> sales_data
sales_data %>%
bin_cols(SALES, bin_type = "value") -> sales_data1
sales_data1
#> # A tibble: 1,000 × 2
#> SALES SALES_va10
#> <int> <int>
#> 1 12210 7
#> 2 14088 9
#> 3 10269 5
#> 4 7003 1
#> 5 7993 2
#> 6 12848 8
#> 7 1686 1
#> 8 6277 1
#> 9 14417 10
#> 10 12804 8
#> # ℹ 990 more rowsNotice that the sum is equal across bins.
sales_data1 %>%
bin_summary() %>%
print(width = Inf)
#> # A tibble: 10 × 14
#> column method n_bins .rank .min .mean .max .count .uniques
#> <chr> <chr> <int> <int> <int> <dbl> <int> <int> <int>
#> 1 SALES equal value 10 10 14417 15882. 19670 65 63
#> 2 SALES equal value 10 9 13365 13879. 14412 70 70
#> 3 SALES equal value 10 8 12404 12829. 13359 78 73
#> 4 SALES equal value 10 7 11637 12011. 12393 83 81
#> 5 SALES equal value 10 6 10864 11249. 11634 88 86
#> 6 SALES equal value 10 5 10124 10467. 10862 96 89
#> 7 SALES equal value 10 4 9300 9677. 10118 103 101
#> 8 SALES equal value 10 3 8498 8919. 9299 111 101
#> 9 SALES equal value 10 2 7226 7920. 8492 126 122
#> 10 SALES equal value 10 1 206 5538. 7225 180 174
#> relative_value .sum .med .sd width
#> <dbl> <int> <dbl> <dbl> <int>
#> 1 100 1032318 15557 1221. 5253
#> 2 87.4 971504 13892. 310. 1047
#> 3 80.8 1000677 12784 269. 955
#> 4 75.6 996891 11979 238. 756
#> 5 70.8 989887 11244. 226. 770
#> 6 65.9 1004841 10439 209. 738
#> 7 60.9 996714 9662 235. 818
#> 8 56.2 990006 8889 240. 801
#> 9 49.9 997945 7938. 359. 1266
#> 10 34.9 996766 5964. 1462. 7019