"NAs introduced by coercion" in seqecreate

Question

I am using seqecreate to specify an event sequence object using TraMineR.

The dataset has 1271 rows and 4 columns, an id, timestamp (timed), and 2 event variables (one numeric and one string). Most of the data have only 1 event per id. I've done other analyses with these data in SAS without issues. When I run seqecreate the procedure stops at id 425 with an error message:

> etv.ev <- seqecreate(id=etvev$id, timestamp=etvev$timed, event=etvev$event2, use.labels=FALSE)

Warning message:

In seqecreate.internal(data = data, id = id, timestamp = timestamp,  :
  NAs introduced by coercion

The offending data in this part of the dataframe look ok (row 535 is the last before the error):

row     id       event2 timed event1
534 2144155702   xETV     0      1
535 2149998801   xETV     0      1
536 2150098301   xETV     0      1
537 2150098301   ETV2    27      2
538 2152463101   xETV     0      1

This error occurs with both event variables. I'd like to understand the problem and how to fix it.

Answer 1

Playing around with your data, I found the problem is with the id column. The error occurs when the id exceeds some value. This is a bug in seqecreate.

A workaround is to pass your ids as a factor:

seqecreate(id=factor(etvev$id), timestamp=etvev$timed, event=etvev$event2)

Hope this helps.