I am creating a Column Expressions node where I pass in both Flow Variables and a table. The variables have names identical to the columns in the table, e.g., “age”, “gender”, and contain a coefficient (number).
I want to create a sum:
sum = 0.0
for (each in columnNames()) {
sum = sum + variable(each)*column(each);
}
This fails with an error that variable(0) does not exist (which is true; variable(“age”) etc. do instead). Using a constant like 1.0 instead works perfectly.
On investigation I found that the columnNames() function does not provide the names, but the indexes instead, contrary to the documentation.
How to get the real column names? I cannot use numerical variable names, and a variableNames() function does not exist either.
You are absolutely correct that the columnNames() function returns an array of column names, like you say.
My mistake was here:
sum = 0.0
for (each in columnNames()) {
sum = sum + variable(each)*column(each);
}
My assumption was that the variable each would be every name in turn (like Python). Turns out this is not the case; it will be the index instead. But the column(each) works fine because you can retrieve the columns using an index, but variable not.
The working code is instead:
sum = 0.0
for (each in columnNames()) {
name = columnNames()[each];
sum = sum + variable(name)*column(name);
}
which is just fine; or perhaps nicer-looking:
sum = 0.0
cols = columnNames()
for (each in cols) {
name = cols[each];
sum = sum + variable(name)*column(name);
}
Many thanks for making me think and for your patience!
