standard deviaition using R

Question

[rsabrina93]

how to perform standard deviation on multiple rows(27482) which contain multiple columns(702) using R? I tried the code below:

code:

for(i in 1:nrow(zscore)){
    print(i)
    SD=sd(as.numeric(zscore[i,]))
    zscore[i,702]=SD
}

Answer 1

[Devon Ryan]

We can use apply and sd functions, see example:

# dummy reproducible data
zscore <- data.frame(matrix(1:10, nrow = 2))
# introduce NA
zscore[1, 2] <- NA
zscore
#   X1 X2 X3 X4 X5
# 1  1 NA  5  7  9
# 2  2  4  6  8 10

# below will give NAs
apply(zscore, 1, sd)
# [1]       NA 3.162278

# we need to remove NAs before getting SD
apply(zscore, 1, sd, na.rm = TRUE)
# [1] 3.415650 3.162278

[rsabrina93]

Does it calculate for the entire data frame?

[Nicolas Rosewick]

yes. The apply function will execute the sd() function on each row separatly (as the second paramter of apply() specifies if the rows (=1) or the columns (=2) will be analyzed.

[rsabrina93]

I get NAs as result for all the rows

[Devon Ryan]

Perhaps you need to zscore = as.numeric(zscore) first or use the na.rm=T option.

[ddiez]

I am assuming Devon's example works fine for you and the problem is when you apply it to your data. Maybe there is a problem with your data not being all numeric, e.g. if you have a data frame with other variables (an so Devon's suggestion about using as.numeric() may or may not work). You could post a few lines of your dataset so that we can take a look at it.