How to reduce the if statement for multiple arguments? - r

Question

I would like to reduce this code:
carro$custo_tprivate = with(carro, ifelse(decile_renda == 1,
renda_fa*0.116,
ifelse(decile_renda == 2, renda_fa*0.106,
ifelse(decile_renda == 3, renda_fa*0.102,
ifelse(decile_renda == 4, renda_fa*0.115,
ifelse(decile_renda == 5, renda_fa*0.124,
ifelse(decile_renda == 6, renda_fa*0.125,
ifelse(decile_renda == 7, renda_fa*0.137,
ifelse(decile_renda == 8, renda_fa*0.141,
ifelse(decile_renda == 9, renda_fa*0.156,
ifelse(decile_renda == 10, renda_fa*0.131, 0)))))))))))
Someone could teach me how to do that?
Thank you very much!

Answer 1

You can technically use match statement to reduce your ifelse statements as below
# put your case data into a data frame
data = data.frame(x= seq(1, 10, by = 1)
, y = runif(10))
# creating your actual data
carro = data.frame(decile_renda = sample(1:10, 10, replace =T)
,renda_fa = runif(10)
)
#Match it to get positions of case statement
pos = with(carro, match(decile_renda, data$x, 0))
# multiply to get results
data$y[pos]*carro$renda_fa

Answer 2

Alternatively, this can be solved by left-joining with a lookup table.
If I understand correctly the nested ifelse() construct, the OP wants to multiply renda with a factor which depends on the value of decile_renda. The factors are given for 10 distinct values of decile_renda. In all other cases, the result must be zero.
Using left join will find matching values of decile_renda in the lookup table. Non-matching rows in carro will get an NA value. These need to be replaced by zero, subsequently.
The lookup table treats the single use cases as data instead of hard-coding. This gives the flexibility to add or change the use cases without changing the code.
Create lookup table and test dataset
# create lookup table
lut <- data.frame(
decile_renda = 1:10,
fa = c(0.116, 0.106, 0.102, 0.115, 0.124, 0.125, 0.137, 0.141, 0.156, 0.131)
)
lut
decile_renda fa
1 1 0.116
2 2 0.106
3 3 0.102
4 4 0.115
5 5 0.124
6 6 0.125
7 7 0.137
8 8 0.141
9 9 0.156
10 10 0.131
# create test dataset
carro <- data.frame(decile_renda = 0:11, renda_fa = 100)
# randomize row order
set.seed(1L) # required for reproducible data
carro <- carro[sample(nrow(carro)), ]
carro
decile_renda renda_fa
9 8 100
4 3 100
7 6 100
1 0 100
2 1 100
5 4 100
3 2 100
8 7 100
6 5 100
11 10 100
12 11 100
10 9 100
Note that the test dataset has been choosen to allow for easy verification of the results.
Base R: merge()
carro <- merge(carro, lut, all.x = TRUE, by = "decile_renda")
carro$custo_tprivate <- with(carro, ifelse(is.na(fa), 0, renda_fa * fa))
carro
decile_renda renda_fa fa custo_tprivate
1 0 100 NA 0.0
2 1 100 0.116 11.6
3 2 100 0.106 10.6
4 3 100 0.102 10.2
5 4 100 0.115 11.5
6 5 100 0.124 12.4
7 6 100 0.125 12.5
8 7 100 0.137 13.7
9 8 100 0.141 14.1
10 9 100 0.156 15.6
11 10 100 0.131 13.1
12 11 100 NA 0.0
Note that the result column custo_tprivate shows 0.0 for the rows with non-matching decile_renda values of 0 and 11 as requested.
However, the drawback here is that merge() does not maintain the original row order (this is why the test dataset uses a random row order for demonstration). Also, the result contains the fa column which is no longer needed.
dplyr
library(dplyr)
carro %>%
left_join(lut, by = "decile_renda") %>%
mutate(custo_tprivate = if_else(is.na(fa), 0, renda_fa * fa)) %>%
select(-fa)
decile_renda renda_fa custo_tprivate
1 8 100 14.1
2 3 100 10.2
3 6 100 12.5
4 0 100 0.0
5 1 100 11.6
6 4 100 11.5
7 2 100 10.6
8 7 100 13.7
9 5 100 12.4
10 10 100 13.1
11 11 100 0.0
12 9 100 15.6
Here, the original row order is kept and the fa column has been removed.
data.table
With data.table we can do an update join where the matching rows of carro are being updated by reference, i.e., without copying the whole object. Only the result column custo_tprivate is appended to carro but not fa which would have to be removed afterwards. Also, for replacing the NA values only the affected rows are updated in place.
This might be an advantage in terms of speed and memory consumption in case of large datasets.
library(data.table)
setDT(carro)[lut, on = .(decile_renda), custo_tprivate := renda_fa * fa]
carro[is.na(custo_tprivate), custo_tprivate := 0]
carro
decile_renda renda_fa custo_tprivate
1: 8 100 14.1
2: 3 100 10.2
3: 6 100 12.5
4: 0 100 0.0
5: 1 100 11.6
6: 4 100 11.5
7: 2 100 10.6
8: 7 100 13.7
9: 5 100 12.4
10: 10 100 13.1
11: 11 100 0.0
12: 9 100 15.6