Run a defined function for various variables in R - r

Question

I have defined a function for descriptive statistics, which is
descriptive_statistics<-function(obj){
n<-length(obj)
cat("mean=" ,mean(obj) , "\n")
cat("s.d.=",sd(obj),"\n")
cat("min=" ,min(obj ) , "\n")
cat("1. Quartil=" ,quantile(obj, 0.25) , "\n")
cat("median=",median(obj),"\n")
cat("3. Quartil=" ,quantile(obj, 0.75) , "\n")
cat ( "max= " , max (obj ) , "\n" )
skew<-(sum((obj-mean(obj))^3)/n)/(sum((obj-mean(obj))^2)/n)^(3/2)
cat("skewness=",skew,"\n")
kurt<-n*sum((obj-mean(obj))^4)/(sum((obj-mean(obj))^2)^2)
cat("kurtosis=",kurt,"\n")
Stat<-n*skew^2/6+n*(kurt - 3)^2/24
}
For any variable that I put in the command, it gives me this output:
descriptive_statistics(data$ewz_2016)
mean= 232024.4
s.d.= 383172
min= 36514
1. Quartil= 81632
median= 121705
3. Quartil= 230405.5
max= 3516100
skewness= 5.930454
kurtosis= 46.42959
My wish would be to put various variables in the command at the same time (e.g. the variables from the columns 3-25) and afterwards copy the table to Excel to format it.
Does anyone have a piece of advice?

Answer 1

I would change the function descriptive_statistics to return the values it computes in the form of a named vector. In what follows the function doesn't print anything, it just computes those values. You can combine the two, yours and this version of it, if you want.
descriptive_statistics2 <- function(obj){
n <- length(obj)
skew<-(sum((obj-mean(obj))^3)/n)/(sum((obj-mean(obj))^2)/n)^(3/2)
kurt <- n*sum((obj-mean(obj))^4)/(sum((obj-mean(obj))^2)^2)
Stat <- n*skew^2/6+n*(kurt - 3)^2/24
c(mean = mean(obj),
s.d. = sd(obj),
min = min(obj),
Quartil = quantile(obj, 0.25), # Note: no need to say it's quartile 1 or 3
median = median(obj),
Quartil = quantile(obj, 0.75), # quantile() will append a 25% or 75%
max = max(obj),
skewness = skew,
kurtosis = kurt,
Stat = Stat
)
}
Now test it with some data.
set.seed(8085) # make it reproducible
n <- 1e2
dat <- as.data.frame(matrix(rnorm(25*n), ncol = 25))
result <- t(sapply(dat[3:25], descriptive_statistics2))
head(result)
# mean s.d. min Quartil.25% median Quartil.75% max
#V3 0.04496632 0.9454383 -2.114690 -0.6352870 0.13869333 0.6789609 1.901803
#V4 0.08136469 1.0594723 -1.955842 -0.6922518 -0.03305246 0.8274921 2.653876
#V5 -0.05899082 1.0225392 -3.031083 -0.6536158 -0.16813752 0.6497042 2.053943
#V6 0.13282456 0.9746959 -1.715005 -0.5399552 0.05815323 0.8383848 2.308354
#V7 0.03935582 1.1629889 -2.669025 -0.6569735 -0.01753087 0.6976987 3.277311
#V8 -0.08203328 1.1111241 -2.755361 -0.9619138 -0.08806534 0.8730580 2.547580
# skewness kurtosis Stat
#V3 -0.20061108 2.414033 2.1014041
#V4 0.32427896 2.455259 2.9890412
#V5 -0.21501670 2.758610 1.0133248
#V6 0.16328790 2.322679 2.3558991
#V7 0.09352129 2.977775 0.1478287
#V8 -0.05664655 2.509671 1.0552417
Finally, to export the result to Excel, use write.csv, if your country uses the period as decimal point or write.csv2 if it uses the comma.
write.csv(result, file = "SCW16.csv", row.names = FALSE)