Difference between minimum match score and minimum candidate score

3924
8
Jump to solution
11-30-2014 11:34 PM
HaniDraidi
Occasional Contributor II

What is the difference between minimum match score and minimum candidate score in Geocoding?

0 Kudos
1 Solution

Accepted Solutions
DavidBlanchard
Esri Contributor

The difference between the two lies in the geocoding process. The address you provide is broken up into multiple representations of that address. For example, "12 Concorde Place, Toronto, ON" could have the following representations:

  • 12 Concorde Place, Toronto, ON
  • Concorde Place, Toronto, ON
  • Toronto, ON
  • ON

Each representation is searched for in the address locator (or in some cases against many locators). Each search produces a list of candidates, each having a match score. This match score represents how closely the search string matched the value in the locator. These candidates are then ranked against each other according to their match score and the accuracy of the locator/representation.

Continuing with the example from above, both "12 Concorde Place, Toronto, ON" and "Toronto, ON" could generate a match score of 100 since the full address was matched to a street address and the city was matched against a point or polygon for the city. However, the street address match is clearly superior and would get a higher candidate score.

This is just one example of what could affect the match and candidate score, but it should help you wrap your head around this obscure concept.

View solution in original post

0 Kudos
8 Replies
JoeBorgione
MVP Emeritus

Exactly what is implied:

In order to be considered a MATCH, the candidate will need score of 'X'

In order to be considered a CANDIDATE, the address will need a score of 'Y'

If you set your MATCH score high, and your CANDIDATE score low, you probably won't get many  matches but you'll probably get a ton of candidates to consider.

That should just about do it....
HaniDraidi
Occasional Contributor II

Joe, thank you very much for your valuable input.

May you please elaborate more about the difference between the MATCH and the CANDIDATE?

0 Kudos
DanPatterson_Retired
MVP Emeritus
0 Kudos
HaniDraidi
Occasional Contributor II

Yes, I checked it out but I did not get the point.

Would you please give me a practical example that can better illustrate the difference?

0 Kudos
DanPatterson_Retired
MVP Emeritus

Did you go through the geocoding tutorial ?  If so, and you still don't get it...try to see if you can geocode your own addresses and see how successful they are.

0 Kudos
HaniDraidi
Occasional Contributor II

Yes, I worked with the tutorial, and built several of address locators, but till now I cannot understand the difference between these two parameters. If you can give a simple and practical example, that will be great.

Thank you Again!

0 Kudos
DavidBlanchard
Esri Contributor

The difference between the two lies in the geocoding process. The address you provide is broken up into multiple representations of that address. For example, "12 Concorde Place, Toronto, ON" could have the following representations:

  • 12 Concorde Place, Toronto, ON
  • Concorde Place, Toronto, ON
  • Toronto, ON
  • ON

Each representation is searched for in the address locator (or in some cases against many locators). Each search produces a list of candidates, each having a match score. This match score represents how closely the search string matched the value in the locator. These candidates are then ranked against each other according to their match score and the accuracy of the locator/representation.

Continuing with the example from above, both "12 Concorde Place, Toronto, ON" and "Toronto, ON" could generate a match score of 100 since the full address was matched to a street address and the city was matched against a point or polygon for the city. However, the street address match is clearly superior and would get a higher candidate score.

This is just one example of what could affect the match and candidate score, but it should help you wrap your head around this obscure concept.

0 Kudos
HaniDraidi
Occasional Contributor II

Thank you very much David, I got the point now

0 Kudos