README.md

In vh-d/hierarchical: Tools for manipulation, transformation and analysis of hierarchical data

hierarchical

hierarchical provides tools for manipulations, transformations and analysis of hierarchical data

Example of decomposition of financial ratios

Simulate random data

library(data.table)
library(data.tree)

# load example data
data("Angell", package = "carData")
DT <- as.data.table(Angell, keep.rownames = TRUE)[sample(1:.N, 20)]
setnames(DT, "rn", "name")

# simulate financial data
DT <- rbindlist(list("2016" = DT, "2007" = DT), idcol = "year")
DT[, cost   := 1000*rgamma(.N, 30, 30)]
DT[, income := 1000*rgamma(.N, 31, 31)]

# compute first differences
setkey(DT, region, name, year)
DT[, cost_diff   := c(NA, diff(cost)),   by = .(region, name)]
DT[, income_diff := c(NA, diff(income)), by = .(region, name)]
DT[, cost_to_income := cost / income]

Convertion of data.table into a data.tree

library(hierarchical)

tree <- dt2tree_ratio(DT, 
                      tree_name       = "Total", 
                      numerator_name  = "cost", 
                      denomiator_name = "income", 
                      dim_names       = c("region", "name"))

Decomposition of ratios

tree$Do(decomp_ratios_root_full, 
        root_denominator_curr = tree$denominator_curr,
        root_denominator_lag  = tree$denominator_lag, 
        multipl = 1)

Output

You can print “ratio_effect”, “weight_effect” and “residual_effect” using data.tree methods:

print(tree, "ratio_curr", "ratio_effect", "weight_effect")
#>                  levelName ratio_curr  ratio_effect weight_effect
#> 1  Total                    1.0242344  0.0187478977  0.0000000000
#> 2   ¦--E                    1.0982822  0.0207443777 -0.0096163751
#> 3   ¦   ¦--Bridgeport       1.1408110  0.0044723195 -0.0082824887
#> 4   ¦   ¦--Erie             0.9413373  0.0198442010 -0.0023120753
#> 5   ¦   ¦--Rochester        1.0772549 -0.0008298147  0.0032020078
#> 6   ¦   ¦--Syracuse         1.2070468 -0.0185743241  0.0174150891
#> 7   ¦   °--Trenton          1.1667929  0.0152941628 -0.0093834525
#> 8   ¦--MW                   0.8845153 -0.0162494611  0.0212309458
#> 9   ¦   ¦--Akron            0.7667057 -0.0001555765  0.0016008515
#> 10  ¦   ¦--Des_Moines       0.8747353 -0.0095667203  0.0048993319
#> 11  ¦   ¦--Flint            1.0057868 -0.0072397696  0.0013945103
#> 12  ¦   ¦--Grand_Rapids     0.8745463 -0.0022124451  0.0106129128
#> 13  ¦   °--Wichita          0.9165926  0.0030851838  0.0026016799
#> 14  ¦--S                    1.1207650  0.0256686538 -0.0086960330
#> 15  ¦   ¦--Atlanta          1.3833759 -0.0071593085  0.0074456291
#> 16  ¦   ¦--Chattanooga      1.1541022  0.0041381951 -0.0063128754
#> 17  ¦   ¦--Jacksonville     1.4078405  0.0207652405 -0.0057685404
#> 18  ¦   ¦--Louisville       0.9915777  0.0045227707 -0.0048828542
#> 19  ¦   ¦--Miami            0.7532886  0.0029089340 -0.0013051727
#> 20  ¦   °--Nashville        1.3178237 -0.0013358667  0.0085001737
#> 21  °--W                    0.9861394 -0.0069512687 -0.0045349363
#> 22      ¦--Denver           0.7459502 -0.0010118923  0.0016424472
#> 23      ¦--Portland_Oregon  1.0993614 -0.0065480976 -0.0026359979
#> 24      ¦--Seattle          1.1212477  0.0011355494 -0.0041835407
#> 25      °--Tacoma           0.9794753  0.0007455685 -0.0005798143

…or print it using ggplot2

library(ggplot2)
plot_tree_ratio(tree)  +
  ggtitle("Decompositon of consolidated cost-to-income ratio") +
  xlab("Entitity") + 
  theme(legend.text = element_text(size = 7))
#> Warning: Removed 6 rows containing missing values (position_stack).

vh-d/hierarchical documentation built on July 5, 2019, 12:23 p.m.