Hello,
I'm running fuzzy matching algorithm for address (please see the sample below) but it won't find create a match score for some obvious addresses (like it won't even find/show for king road despite they are almost identical) . Also, I would like to see the unique ID that I created, in the output? Any chances?
Also,If I create one more column with name field, can i see them in output?
Name1 | Address | Name2 | Address2 | MatchScore | MatchScore_Address | MatchKey |
John Smith | 1234 Main Street, Pittsburgh, PA | John Smith | 1234 Main Street, Pittsburgh, PA | 100 | 100 | 1234 |
John Smith | 1235 Main Street, Pitts, PA | Daniel Watson | 1235 Main Street, Pittsburgh, PA | 87 | 87 | 1235 |
This is all the addreses where I use fuzzy match
1234 Main Street, Pittsburgh, PA |
1235 Main Street, Pittsburgh, PA |
1234 Main Street, Pittsburgh, PA |
1235 Main Street, Pitts, PA |
1234 King Rd, Houston,TX |
1235 King Rd, Houston,TX |
Solved! Go to Solution.
Fuzzy matching can be challenging. Regarding your question of the streets not matching, the first thing that comes to mind is that your key is too short to include the street name. If it's only matching numbers, the keys would not match. I've had that happen before with the opposite result, where it matched 114 Main and 114 Princeton because it was only picking up the 114 portion.
There is an option in the fuzzy match tool configuration to output the match key, so just make sure that option is selected.
Here's a really great 10 minute video by Alteryx that helped me a lot:
https://www.youtube.com/watch?v=JlZiM2YSCCY
If you want to dive deeper, Cailin Swingle's webinars are great! Check out a previously recorded live training (look for Fuzzy Matching):
http://www.alteryx.com/virtual-training#virtualprev
Just to make sure you are aware of this, there is a CASS tool that will standardize US addresses to USPS standards. Granted you need to have a license for the spatial data in Alteryx, but when it comes to working with addresses, "best practice" is to have them all standardized first...this avoids any need for "fuzziness".
Update on the fuzzy matching videos:
Current videos can be found here http://www.alteryx.com/virtual-training
More specifically, the video I was referring to is here: