Confusing non-numeric argument to binary operator error - r

Question

I cannot seem to even create a reproducible example on this as it works fine when I go through the code one line at a time.
The error message I get is as follows:
"Error in testData[, colCheck][length(testData[, colCheck])] - testData[, :
non-numeric argument to binary operator "
Both colCheck and testData$linearcorrd15N are numeric and like I said, the calculation works fine when I run it at that line. The error comes only when I run the function from QTest(df, colCheck).
Here is an example of what some of the code looks like. It will not produce an error, but maybe you can see something that I don't.
QTest <- function(testData, colCheck)
#%#
# testData <- This is the entire data frame for the std/ref that has too high
# of a SD, this way the data frame can be returned without the outlier
# colCheck <- The column name for values that were flagged for having too high of a SD
# This Q test info provided by: https://www.statisticshowto.com/dixons-q-test/
#%#
{
#Get the mean of the highest and lowest values
testData <- arrange(testData, desc(testData[, colCheck]))
len <- length(testData[,colCheck])-1
high <- sapply(1:len, function(i) testData[,colCheck][i])
meanhigh <- mean(high)
testData <- arrange(testData, (testData[, colCheck]))
low <- sapply(1:len, function(i) testData[,colCheck][i])
meanlow <- mean(low)
#If the mean of the lowest numbers is lower than the mean of the highest numbers, do this
if(meanlow < meanhigh){
QexpVal <- abs((testData[, colCheck][2] - testData[, colCheck][1])/
(testData[, colCheck][length(testData[, colCheck])] - testData[, colCheck][1]))
outlier <- testData[,colCheck][1]
closest <- testData[,colCheck][2]
#else if the mean of the lowest numbers is higher than the mean of the highest numbers, do this
} else {
QexpVal <- abs((testData[, colCheck][length(testData[,colCheck])-1] - (testData[, colCheck][length(testData[,colCheck])])) /
(testData[,colCheck][length(testData[,colCheck])]) - (testData[,colCheck][1]))
outlier <- testData[,colCheck][length(testData[,colCheck])]
closest <- testData[,colCheck][length(testData[,colCheck])-1]
}
return(QexpVal)
}
df <- data.frame(Row = c(1, 2, 3, 4, 5), Identifier.2 = "36-UWSIF-UT Glut1", linearcorrd15N = c(-11.63433,
-22.13869, -57.21795, -17.06438, -16.23358))
colCheck <- as.numeric(grep("linearcorrd15N", colnames(std1)))
QTestCorrVals <- QTest(df, colCheck)

Answer 1

It seems you realy overcomplicate this function by pushing the whole table in the function and loop over everything and read a value again from the whole table...
just the part to get meanhigh and meanlow requires this:
v <- df[, colCheck]
v <- v[order(v)]
n <- length(v)
meanhigh <- mean(v[2:n])
meanlow <- mean(v[1:n-1])
Or if you use the decreasing ordering this:
v <- df[, colCheck]
v <- v[order(v, decreasing = T)]
n <- length(v)
meanhigh <- mean(v[1:n-1])
meanlow <- mean(v[2:n])
Full function
Hereby the full code using this approach and I agree that is not the specific question you asked, but the way you coded it is highly inefficient and error prone by every time take the whole data.frame and subset it and recalculate lengths every time. Also you just have to order once, as if the lowest is on top, the highest is per definition on the bottom. Then play around with the 1 for first and 2 for second and n for last and n-1 for second last.
df <- data.frame(Row = c(1, 2, 3, 4, 5), Identifier.2 = "36-UWSIF-UT Glut1", linearcorrd15N = c(-11.63433,
-22.13869, -57.21795, -17.06438, -16.23358))
colCheck <- as.numeric(grep("linearcorrd15N", colnames(df)))
QTest <- function(v) {
v <- v[order(v)]
n <- length(v)
meanhigh <- mean(v[2:n])
meanlow <- mean(v[1:n-1])
if(meanlow < meanhigh) {
QexpVal <- abs((v[2]-v[1])/(v[n]-v[1]))
outlier <- v[1]
closest <- v[2]
} else {
QexpVal <- abs((v[n-1]-v[n])/(v[n]-v[1]))
outlier <- v[n]
closest <- v[n-1]
}
return(QexpVal)
}
QTestCorrVals <- QTest(df[, colCheck])
Side note
Using the column index number works slightly different whether your data is a data.frame or a data.table
class(df)
df[, colCheck]
dt <- data.table(df)
class(dt)
dt[, ..colCheck]
dt[, colCheck] # throws an error