daudi / epidata-xml-to-r Goto Github PK

Hello,
Thank you for your code.
I am a statistician and a R beginner. I need to import multiple files .epx from Epidata and your code would be very usefull.
Unfortunatly, error "Error in xmlChildren(these.data)[[i]] : subscript out of bounds" is generated when I run your code. Any idea to help me?
Thank you again for your code and your help,
Laura

Hi David,
Logical type is not well documented

Field type = 0
Can have values "Y", "N" or be blank (system missing); value labels not allowed as far as I know.

conversion code:

} else if (fld.type == 0){
## Logical - empty to NA, Y to TRUE, else to FALSE
x[x==""] <- NA
x <- x=="Y"

Jamie

It is possible to enter data, and later add labels that do not include all of the values entered previously. Importing such a file causes the application of labels to fail because the factor levels don't match.

Detect it, report it, and attempt to apply the labels where possible.

From Jamie: Perhaps give a warning and assign value label such as "Unknown label 1" so that every value would have a corresponding unique label.

But might be better to leave the data originally entered, as this would allow the user to know what the values were and recode manually in R.

Some .epx files will have no value labels. Catch this in get.epidata.value.labels with:

y <- xmlElementsByTagName(x, "ValueLabelSet", rec = TRUE)
if (xmlSize(y)==0) return(NULL)

I think it should be possible to use handlers in xmlTreeParse() to filter rows when parsing the epidata XML file. This could be useful if the file is very large and there is a meaningful way of using a subset (perhaps random selection, or rows that match certain criteria).