I've created a multi-role address locator that includes point addresses, street centerlines, parcels, POIs, and reference tables for alternate spellings and similar variations. The locator works great when I manually search using the Locate tool, but I'm having trouble getting it to work for batch geocoding.
I'm trying to geocode a table in CSV format that contains simple addresses and intersections. Unfortunately, both types of data are failing to geocode properly. The dataset I'm working with has over 200,000 records, but the best match rate I've achieved so far is just 12%. ( I have tested manually looking up the unmatched ones and I am getting actual matches)
I also am only seeing a match score that is either 100 or 0 and nothing in between
Am I missing something obvious here? Any insights or suggestions would be greatly appreciated!
Thanks in advance for your help.
As a test I would create simpler address locators with one role and see how that works compared to multi-role. If that works as expected for each role, then I would start to add more than 1 role to each locator to see if you still get good results. Good luck.
I've tested smaller datasets and noticed something odd. I understand that altering the format of the input data might change the results, but here's the strange part: when an address fails to match during batch geocoding, I can copy and paste that exact same address into the Locate tool, and it pops up immediately without any format changes or edits. But when processed in a batch, it returns a 0 match score.
The goal is to avoid having to modify the input data, as it's being regularly pulled from another software and needs to be geocoded immediately. Constantly cleaning or reformatting the data isn't a feasible solution.
Does anyone have insights into why this discrepancy might occur between manual and batch geocoding? Could there be something in the batch process that's affecting how the locator reads or interprets the data? Any suggestions on how to "address" this would be greatly appreciated!