When using the ChangeDetector to compare attribute values, decimal and datetime values create false-positives due to precision (i.e. 10.00 vs. 10.0001). Perhaps comparison of decimal values could be done via abs(value1 - value2) > epsilon. The user could enter the epsilon value. Dates vs. DateTimes could be handled similarly.
+3- Contributor
Was just pulling my hair out with this today. It wasn’t a matter of close precision like the post but I inserted into SQL Server from an Oracle dataset and because of differences in how the data was represented by each vendor, it wouldn’t match. 25.8 <> 25.80000000 according to ChangeDetector but was flagged as a duplicate by DuplicateFilter.
I ended up using a regex replace in an attribute manager to trim trailing 0s, but more control over decimal and datetime matching is worth an upvote!
Cheers, Ken
1 person upvoted this
Reply
Helpful Members This Week
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.