2

enter image description here

I am trying to merge two datasets of golfcourse data with almost similar information. my end goal is to have all the data into one csv file that have merged information from both csv files. I have tried using feature merger but it does not merge my attributes or columns together based on name or address field as a key.

Can someone take a look at the picture below and see if i am using the transformers right or any other way for me to combine these data sets into one having no duplicates of rows and fields.

  • Can you expand on what you mean by "almost similar information"? On what basis are you trying to merge the data? What error does FeatureMerger give? – mKurowsKi Jul 14 '15 at 22:01
  • almost similar because of naming convention found in the address like some would spell "blvd" - "boulevard" - "bvd" and so on. – Gonzalo68 Jul 14 '15 at 22:10
  • I am trying to merge based on address or name. The featuremerger does not merge anything in my output and still remain as separate fields. – Gonzalo68 Jul 14 '15 at 22:10
  • Are you able to edit your post to have a couple of examples of records that you would expect to match? Also the settings that you have for the Matcher. I would expect to see to matchers to handle the 'OR' condition, one for address and one for name. – MickyT Jul 15 '15 at 2:37
  • I think you're okay up to the AttributeKeeper. I can't see a FeatureMerger in the screenshot above but if you need to merge records you need key fields with the same value in each, i.e the Addresses must match exactly. Looking at your fields, you may be best attempting it with a combination of Street, City, State and Zip as they are easily cleansed and it is, I'm assuming given their size, unlikely that you'll get different Golf Courses with the same values. – James4571 Jul 15 '15 at 13:06
1

Note: You don't need the FeatureHolders that you have in the above; they accomplish nothing, although they probably aren't hurting either in this case.

While you will eventually be able to get your solution to work, it will be a royal pain. Address data is about as inconsistent as it gets as you're discovering. i.e. Is "st" for "street" or "Saint"? How about hyphens, some datasets will use them, some won't. Etc etc.

Instead I'd propose a different solution - pass all the addresses through a geocoder, and then use the coordinates that come back as the unique ID which you can match by.

FME includes several GeoCoding transformers, but if you don't want to go that route you can do that separately in another package, other questions may help: Bulk Geocode 20 million records and Geocoding USA addresses that cannot be sent over internet? for instance

This way you're giving the address problem to other people who have spent years and millions of $ trying to fix it rather than doing it yourself.

This solution won't be perfect either, but it should hopefully help as a follow-up round from your above solution.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.