Street Geocoder Confidence Score - How Calculated?

[Morgan_Kraman]

The definition of "Score" on the Alteryx website for the Street Geocoder tool doesn't provide much detail into how this score is determined. I found a Confidence Score definition for OpenCage and Precisely (page 61 - this also uses TomTom data in addition to TIGER and NavTEQ) that is more detailed, and I'm guessing is probably the same general process that the Alteryx tool is using, but is it possible to obtain a similar definition to the examples provided specifically for the Alteryx tools? Is Alteryx assigning the score or is it using a TomTom algorithm that the Alteryx tool is just passing through?

[CharlieS]

That's a good question; I'm not aware of any more verbose documentation on that score. I've never paid too much attention to the exact score value and only ever utilized the [GeoLevel] values like "Street" or "ZIP+4" to sort the results. 

Let's see if any of the other significantly spatial ACEs have additional insights:

@Deanna @Hollingsworth @Samanthaj_hughes 

[Hollingsworth]

Same here @CharlieS. I think this is a question for the Alteryx data folks.

[Morgan_Kraman · Accepted solution]

Thank you @CharlieS and @Hollingsworth for the quick reply! I submitted a ticket and the response I received is that this data comes from TomTom's Geocoder. Each of the elements in a request is matched to their index and contributes to the score, but it is not clear how this number is generated other than via an algorithm. The result scores are relative to each other, not a normalized scale, and the highest number (100) is the best result. In other words, ‘result 1 is 2x more likely to be relevant than result 2, if the score is twice as high’. However it is not possible over all queries to generalize the meaning of this confidence value. E.g. you cannot generally say ‘anything below 2 is irrelevant’.