How to state na.rm=T for functions?

[xiekunwhy]

Hi,
I think NA in data is the most common case in practice, and I don't known how to specify na.rm=T or na.rm=F in some NA sensitive functions like mean sum?

Xie Kun.

[tavareshugo]

Hi @xiekunwhy

Thanks for your interest in the package, and you're absolutely right, NA is ubiquitous in data!

With the way winScan() works at the moment, you have two options: 1) remove the missing values before doing the window summaries or 2) create a custom function to pass to winScan().

I will exemplify both with some simple example data:

library(WindowScanR) # load the library

# Create some example data with missing values
df <- data.frame(x = 1:10, y = c(1, 1, 2, NA, 3, 3, NA, 3, 2, 2))

This is what it looks like:

   x  y
   1  1
   2  1
   3  2
   4 NA
   5  3
   6  3
   7 NA
   8  3
   9  2
  10  2

---

Option 1
Remove missing values before doing the window summaries. For example:

df_filtered <- subset(df, !is.na(y))

Which now looks like:

   x y
   1 1
   2 1
   3 2
   5 3
   6 3
   8 3
   9 2
  10 2

So you can now do:

winScan(df_filtered, position = "x", values = "y", win_size = 2, funs = c("mean", "sum"))

---

Option 2
Create a custom function

Instead, we can make custom functions of mean() and sum() with different defaults, for example:

mean_na_rm <- function(x, ...){
  mean(x, na.rm = TRUE, ...)
}

sum_na_rm <- function(x, ...){
  sum(x, na.rm = TRUE, ...)
}

And now we use these functions in our winScan() call with the original data:

winScan(df, position = "x", values = "y", win_size = 2, funs = c("mean_na_rm", "sum_na_rm"))

Hope this helps!

---

ps - for future reference, it helps if you provide with a reproducible example. Thanks!