calculate weights from a column in a dataframe with conditions R-CodePudding

I need to calculate weights from a column in a dataframe based on some conditions.

I have Total Assets from several banks, with different countries, years and specializations.

For each bank I want to calculate a weight (w) where w(i) = Tot_Asset (bank) / sum (Tot_Ass of all banks within same year, country and Specialization)

Example dataframe:

banks <- c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
Country <- c("NL", "ES", "IT", "IT", "ES", "NL", "FR", "NL", "ES", "NL", "IT", "IT", "NL", "ES", "NL")
year <- c(2020, 2019, 2018, 2019, 2020, 2020, 2018, 2019, 2019, 2019, 2018, 2019, 2020, 2018, 2020)
Specialization <- c("cooperative", "saving", "cooperative", "cooperative", "saving", "cooperative", "saving", "cooperative", "cooperative", "saving", "cooperative", "saving", "cooperative", "cooperative", "cooperative")
Tot_Assets <- c(100, 200, 145, 300, 200, 345, 543, 190, 150, 120, 310, 210, 110, 210, 220)
data <- data.frame(banks, Country, year, Specialization, Tot_Assets)

As an example of what I would like to obtain:

The bank 1 is in NL, is a cooperative and the Total Asset is from 2020; bank 6, 13 and 15 has the same caracteristics. So the program has to do :

w(1) = 100 / (100 345 110 220) = 0.13
w(6) = 345 / (100 345 110 220) = 0.45
and so on

I hope that I explained myself and that you can help me, thanks in advance!

CodePudding user response：

We may group by 'Country', 'year', 'Specialization', and create the 'weights' column by dividing the 'Tot_Assets' with the sum of 'Tot_Assets'

library(dplyr)
data <- data %>% 
  group_by(Country, year, Specialization) %>%
  mutate(weights = Tot_Assets/sum(Tot_Assets)) %>%
  ungroup

-output

data
# A tibble: 15 × 6
   banks Country  year Specialization Tot_Assets weights
   <dbl> <chr>   <dbl> <chr>               <dbl>   <dbl>
 1     1 NL       2020 cooperative           100   0.129
 2     2 ES       2019 saving                200   1    
 3     3 IT       2018 cooperative           145   0.319
 4     4 IT       2019 cooperative           300   1    
 5     5 ES       2020 saving                200   1    
 6     6 NL       2020 cooperative           345   0.445
 7     7 FR       2018 saving                543   1    
 8     8 NL       2019 cooperative           190   1    
 9     9 ES       2019 cooperative           150   1    
10    10 NL       2019 saving                120   1    
11    11 IT       2018 cooperative           310   0.681
12    12 IT       2019 saving                210   1    
13    13 NL       2020 cooperative           110   0.142
14    14 ES       2018 cooperative           210   1    
15    15 NL       2020 cooperative           220   0.284

CodePudding user response：

A base R option using ave (for grouping purpose) and proportions (to calculate the weight distribution)

transform(
  data,
  weights = ave(Tot_Assets,Country,year,Specialization, FUN = proportions)
)

gives

   banks Country year Specialization Tot_Assets   weights
1      1      NL 2020    cooperative        100 0.1290323
2      2      ES 2019         saving        200 1.0000000
3      3      IT 2018    cooperative        145 0.3186813
4      4      IT 2019    cooperative        300 1.0000000
5      5      ES 2020         saving        200 1.0000000
6      6      NL 2020    cooperative        345 0.4451613
7      7      FR 2018         saving        543 1.0000000
8      8      NL 2019    cooperative        190 1.0000000
9      9      ES 2019    cooperative        150 1.0000000
10    10      NL 2019         saving        120 1.0000000
11    11      IT 2018    cooperative        310 0.6813187
12    12      IT 2019         saving        210 1.0000000
13    13      NL 2020    cooperative        110 0.1419355
14    14      ES 2018    cooperative        210 1.0000000
15    15      NL 2020    cooperative        220 0.2838710