Data manipulate: how to get smallest level of two attributes

Hi,

I have a table A of points data has 4 columns

ID longitude latitude location

1 117 56 1155

2 117 57 1155

3 118 56 1155

4 118 57 1156

5 119 56 1156

6 117 59 1156

ID column has unique value. Longitude and latitude repeats. location means the ID data located into some polygons.

How can I shoose from the table A but to make sure: 1) at least one point (ID) from each 'location' value. More than one is also acceptable. 2) the chosen data's lon and lat should have the smallest number of records (levels). If talbe B is the points I choose:

unique(B(:,longitude))*unique(B(:,latitude))

is the smallest. (It doesn't matter if the unique(B(:longitude)) or the unique(B(:latitude)) is the smallest.