Count each customer once in this query - mysql

Question

I have two tables: one is a list of store locations (with lat/long) and the other is a customer list (with address lat/long). What I need is a query that shows how many customers are within certain ranges from each store. The goal is to have each customer counted once in the the distance range that is closest to a store. That is, each customer should only be counted once. For example, if they are 2 miles from one store and 5 from another, then only count them as being associated with the first store.
The query below is supposed to roll all this up so basically I can see the maximum distance all customers are from any store.
This is what my query looks like:
SELECT CASE
WHEN dist < 8046. THEN 1
WHEN dist < 16093. THEN 2
WHEN dist < 40233. THEN 3
WHEN dist < 80467. THEN 4
WHEN dist < 160934. THEN 5
END AS grp,count(*)
FROM (SELECT s.id, s.identifier, ST_Distance_Sphere(s.the_geom, c.the_geom) AS dist FROM full_data_for_testing_deid_2 c, demo_locations_table s)
AS loc_dist
GROUP BY grp
And here's the result:
| Count | grp |
|---------|------|
| 2860 | 1 |
| 4858 | 2 |
| 12735 | 3 |
| 11432 | 4 |
| 23950 | 5 |
| 1002970 | null |
There are only 32048 customers in my database, so this isn't quite working right. If it were, I'd expect the values to increase linearly, but in my results there are more customers in group 3 v. 4, which shouldn't be the case. In addition, groups 1-5 should add up to 32048, as each customer should only be counted once.
Any thoughts on how to adjust this such that each customer is only counted once?

Answer 1

To count each customer only once (in Postgres 9.3+):
SELECT CASE
WHEN s.dist < 8046.0 THEN 1
WHEN s.dist < 16093.0 THEN 2
WHEN s.dist < 40233.0 THEN 3
WHEN s.dist < 80467.0 THEN 4
WHEN s.dist < 1609340.0 THEN 5
END AS grp
, count(*)
FROM full_data_for_testing_deid_2 c
, LATERAL (
SELECT s.id, s.identifier, ST_Distance_Sphere(s.the_geom, c.the_geom) AS dist
FROM demo_locations_table s
ORDER BY dist
LIMIT 1
) s
GROUP BY 1;
This takes every customer exactly once and finds the closest location to go with it before aggregating.
But I don't think ST_Distance_Sphere() uses a GiST index on the_geom.
Consider ST_DWithin() instead if performance is an issue.
How to alter this PostGIS ST_distance_sphere query to give the answer for all points in the table, not just one?