topic Re: Proc geocode issues in SAS Procedures

Proc geocode issues

rprincey — Tue, 28 Nov 2017 14:35:23 GMT

Hi,

I've run proc geocode on a dataset containing address, city, state, zip but it only returned latitudes in the high negatives (ie, down in antartica). The code ran successfully and showed a high percentage of matches at the street level, but the latitudes are wrong.

Is there a known bug with the procedure?

Thanks!

16 proc geocode /* Invoke geocoding procedure */
17 method=street /* Specify geocoding method */
18 data=members_transz /* Input data set of addresses */
19 out=members_geo /* Output data set with X/Y values */
20 addressvar=address2
21 addresszipvar=zip
22 addressstatevar=state
23 addresscityvar=city
24 lookupstreet=lookup.usm
25 ;
26 run;

NOTE: Address data set BOB.IMAX_MEMBERS_DRVS_TRANSZ has 24,957,395 observations.

_________________ Geocoding Progress _____________________
_____ Completed ____ _____________ Time _______________
Percent Obs Current Remaining Completion
0% 0 15:07:31 -- --
1% 249,574 15:07:45 00:23:29 15:31:15
2% 499,148 15:08:00 00:23:29 15:31:29
3% 748,722 15:15:36 04:21:28 19:37:04
4% 998,296 15:21:51 05:43:53 21:05:44
5% 1,247,870 15:25:31 05:41:53 21:07:24
10% 2,495,740 15:37:07 04:26:21 20:03:28
15% 3,743,610 15:46:22 03:40:11 19:26:34
20% 4,991,480 15:55:05 03:10:17 19:05:23
25% 6,239,349 16:03:33 02:48:07 18:51:41
30% 7,487,219 16:11:57 02:30:19 18:42:17
35% 8,735,089 16:20:14 02:15:03 18:35:18
40% 9,982,959 16:28:31 02:01:30 18:30:02
45% 11,230,828 16:36:49 01:49:08 18:25:58
50% 12,478,698 16:45:02 01:37:31 18:22:33
55% 13,726,568 16:53:18 01:26:33 18:19:52
60% 14,974,438 17:01:36 01:16:03 18:17:40
65% 16,222,307 17:09:57 01:05:55 18:15:53
70% 17,470,177 17:18:17 00:56:02 18:14:19
75% 18,718,047 17:26:37 00:46:21 18:12:59
80% 19,965,917 17:34:57 00:36:51 18:11:49
85% 21,213,786 17:43:13 00:27:28 18:10:42
90% 22,461,656 17:51:30 00:18:13 18:09:43
95% 23,709,526 17:59:43 00:09:03 18:08:47
100% 24,957,395 18:07:44 00:00:00 --

_________ Geocoding Summary _____________________________
Address data: BOB.IMAX_MEMBERS_DRVS_TRANSZ
Output data: WORK.IMAX_MEMBERS_DRVS_GEO
STREET lookup data: LOOKUP.USM
CITY lookup data: MAPSGFK.USCITY_ALL
ZIP lookup data: SASHELP.ZIPCODE
Geocoding method: Street level
Run date: 27Nov2017
Obs processed: 24,957,395
Elapsed time: 03:00:13
Obs per minute: 138,477
Street matches: 21,568,915
ZIP matches: 3,324,278
City matches: 0
Not matched: 64,202
_________

Re: Proc geocode issues

ballardw — Tue, 28 Nov 2017 15:21:19 GMT

I would say the most likely place to look is at the values in you lookup dataset. Your lookup set may be projected for a specific use and not returning actual latitude and longitude values.

Re: Proc geocode issues

Reeza — Tue, 28 Nov 2017 16:19:29 GMT

Not that I'm aware of...have you checked what method was used for those particular records, ie the match/lookup type I believe it's called..

Is it possible to post a few addresses - or similar type addresses that we can replicate the issue with?

And what version of SAS are you using and did you update your maps data from here;

http://support.sas.com/rnd/datavisualization/mapsonline/html/geocode.html

Re: Proc geocode issues

rprincey — Tue, 28 Nov 2017 20:18:14 GMT

Thanks for the replies - the issue is resolved.

Re: Proc geocode issues

Reeza — Tue, 28 Nov 2017 20:44:37 GMT

@rprincey What was the issue?

Re: Proc geocode issues

rprincey — Tue, 28 Nov 2017 20:48:43 GMT

I unfortunately don't know. When I first put the data in map (tableau), it all showed up in Antartica.
Then I re-ran it and it worked.
The one thing that may have caused the problem was that, in a data step, I originally copied the lat/long field to new variables for naming sake:
latitude=x;
longitude=y;
Maybe that caused the issue?

Re: Proc geocode issues

ballardw — Tue, 28 Nov 2017 23:47:03 GMT

@rprincey wrote:
I unfortunately don't know. When I first put the data in map (tableau), it all showed up in Antartica.
Then I re-ran it and it worked.
The one thing that may have caused the problem was that, in a data step, I originally copied the lat/long field to new variables for naming sake:
latitude=x;
longitude=y;
Maybe that caused the issue?

Very likely, longitude is the line that runs north/south on a map and the value is used for the direction east/west: i.e. X coordinate in mapping by default. So you transposed the values from east/west to north south. You should have had a largish number of points not map at all as the values from -90 to -180 wouldn't have a usable coordinate pair for mapping.

If you look in the Sashelp.zipcode file you'll notice that the variable named X is labeled Longitude.