TheSamba.com :: View topic - Distance search not working in classifieds correctly

Tue Dec 06, 2011 7:36 am

Good morning, maybe I'm nuts, but I like to search in a 250 mile radius from my zip 40165. I almost always get hits for ads in South Carolina, this morning I have an ad from Easley, SC 29642 which is 432 miles away according to google maps. Am I doing it wrong? I also sort the results by date placed, not sure if that helps or hurts.

PS still an awesome site !

Tue Dec 06, 2011 9:13 am

The sorting would not affect anything.

It does seem like the 250 mile range goes too far. I see some ads from Southern UT from my Phoenix zip code, which is further than 250 miles from here. I will check it out.

Tue Dec 06, 2011 1:42 pm

EverettB, I was thinking in the SQL code it was trumping the distance where clause with the ranking by date, but I wasn't sure, just a thought :) - Thanks

Wed Dec 07, 2011 7:12 pm

My saved favorite search of 100 miles from Park City, Utah (just outside Salt Lake) now includes Arizona and beyond.

Wed Dec 07, 2011 9:13 pm

Perhaps the US is shrinking, that would be kinda good. Living in Kentucky I've always wanted to see California, soon it will be about 2-3 hours away :)

Wed Dec 07, 2011 10:13 pm

I was tweaking some stuff today and it may have corrected some of the distance from zip searching.

This includes up to about 10 minutes ago.

Please try again and let me know if it still seems weird.

Include your zip code so I can try it out, although I might not know exactly what looks weird from your zip.

Thu Dec 08, 2011 7:02 am

EverettB, just gave it a shot, plugged in my zip of 40165 - 250 miles from me, on the left I picked early buses, vanagons & type 3's. A few picks down is a square that's listed in South Carolina, zip code was 29651 which is 437 miles away.

Thu Dec 08, 2011 9:42 am

Ok, I added some more math to it so now it should exclude those.

Note that now less ads are returned, which should be correct, but it will more difficult to know if something is missing rather than seeing it included.

Thu Dec 08, 2011 10:03 am

It's getting better, the furthers I see now is a Asheville, NC 28806 which is about 360 miles away, that's with a search of 250 miles. Does the forum software allow you to manipulate the SQL used in the search?

Thu Dec 08, 2011 10:31 am

Yes, I am changing it when I was trying things, as well as adding to and changing the math it is using.

Google search for those zips:
http://maps.google.com/maps?q=40165+to+28806+&...=m&z=8

Google says 325 miles but I think that is closer to 250 if you take a straight shot there, no roads.

More of a straight shot is 308 miles:
http://maps.google.com/maps?saddr=40165&daddr=...=m&z=8

Also my data is not from Google so it could be slightly different.

Thu Dec 08, 2011 10:32 am

EverettB wrote: Yes, I am changing it when I was trying things, as well as adding to and changing the math it is using.

Google search for those zips:
http://maps.google.com/maps?q=40165+to+28806+&...=m&z=8

Google says 325 miles but I think that is close to 250 if you take a straight shot there, no roads.

More of a straight shot is 308 miles:
http://maps.google.com/maps?saddr=40165&daddr=...=m&z=8

Also my data is not from Google so it could be slightly different.

Perhaps Google is searching by radius rather than an assumed driven path?

Thu Dec 08, 2011 10:33 am

Google is doing it via the roads.

I'm calculating it it as straight shot there, while factoring in the curvature of the earth.

The zip code/latitude/longitude data I have is only to 2 decimal places so there is going to be a higher degree of variance that other data that may be available to Google, etc.

Fri Dec 09, 2011 10:26 am

As part of these changes, I also added a few more "distance from zip" selections from the drop-down.

I added 300 & 400 miles so now the complete list is:
10
25
50
75
100
200
250
300
400
500

250 is sort of superfluous now but I left it there anyway.

Tue Dec 13, 2011 3:50 pm

EverettB wrote: Google is doing it via the roads.

I'm calculating it it as straight shot there, while factoring in the curvature of the earth.

The zip code/latitude/longitude data I have is only to 2 decimal places so there is going to be a higher degree of variance that other data that may be available to Google, etc.

Can you adjust for "By Car" or "By Boat" option for the distance?

Case in point: Wisconsin by Car and Wisconsin by Boat. There are Oshkosh items coming up as being 200 miles away. :lol:

Tue Dec 13, 2011 4:00 pm

EverettB wrote:
I'm calculating it it as straight shot there, while factoring in the curvature of the earth.

Assuming Earth as a sphere or as a flattened spheroid? If the latter, do the values represent the shortest or longest distance?

Tue Dec 13, 2011 8:41 pm

The Sage wrote: EverettB wrote: Google is doing it via the roads.

I'm calculating it it as straight shot there, while factoring in the curvature of the earth.

The zip code/latitude/longitude data I have is only to 2 decimal places so there is going to be a higher degree of variance that other data that may be available to Google, etc.

Can you adjust for "By Car" or "By Boat" option for the distance?

Case in point: Wisconsin by Car and Wisconsin by Boat. There are Oshkosh items coming up as being 200 miles away. :lol:

Beetles float so you are set either way. ;)

Tue Dec 13, 2011 8:43 pm

mattcuddy wrote: EverettB wrote:
I'm calculating it it as straight shot there, while factoring in the curvature of the earth.

Assuming Earth as a sphere or as a flattened spheroid? If the latter, do the values represent the shortest or longest distance?

Sphere but it runs off the latitude so the sphere's radius varies.

I'm not sure if that answers your question though.
It's the distance along the curve from one latitude/longitude to another latitude/longitude.

Wed Dec 14, 2011 11:07 am

EverettB wrote: Sphere but it runs off the latitude so the sphere's radius varies.

The sphere's radius can only vary if you're assuming its not actually a sphere. ;)

EverettB wrote:
I'm not sure if that answers your question though.
It's the distance along the curve from one latitude/longitude to another latitude/longitude.

I'm guessing you're assuming the Earth is a perfectly round sphere (radius is same at every point), and the distance is calculated as the path along the Earth's surface created by a plane that goes through the start point, the end point and the earth 's center.

I used to work for the DoD doing long geodesics on the Earth for missile guidance stuff so I was just curious as this stuff doesn't come up in everyday life too often. But I'll stop posting now before I totally publicly nerd out on the internet. :)

Wed Dec 14, 2011 11:36 am

mattcuddy wrote: EverettB wrote: Sphere but it runs off the latitude so the sphere's radius varies.

The sphere's radius can only vary if you're assuming its not actually a sphere. ;)

EverettB wrote:
I'm not sure if that answers your question though.
It's the distance along the curve from one latitude/longitude to another latitude/longitude.

I'm guessing you're assuming the Earth is a perfectly round sphere (radius is same at every point), and the distance is calculated as the path along the Earth's surface created by a plane that goes through the start point, the end point and the earth 's center.

Yes, I was confusing myself by looking at the math. What you posted is correct.

It is using the mean radius.