The SQL group-by operator plays an important role in summarizing and aggregating large datasets in a data analytics stack. The Similarity SQL-based Group-By operator (SGB, for short) extends the semantics of the standard SQL Group-by by grouping data with similar but not necessarily equal values. While existing similarity-based grouping operators efficiently realize these approximate semantics, they primarily focus on one-dimensional attributes and treat multi-dimensional attributes independently. However, correlated attributes, such as in spatial data, are processed independently, and hence, groups in the multi-dimensional space are not detected properly. To address this problem, we introduce two new SGB operators for multi-dimensional data. We implement and test the new SGB operators and their algorithms inside PostgreSQL. The overhead introduced by these operators proves to be minimal and the execution times are comparable to those of the standard Group-by. The experimental study, based on TPC-H and a social check-in data, demonstrates that the proposed algorithms can achieve up to three orders of magnitude enhancement in performance over baseline methods developed to solve the same problem.

##Documentation Similarity Group-by Operators for Multi-dimensional Relational Data Mingjie Tang, Ruby Y. Tahboub, Walid G. Aref, Mikhail Atallah, Qutaibah Malluhi, Mourad Ouzzani, Yasin. Siva in IEEE Transactions on Knowledge and Data Engineering (TKDE2015)

http://www.postgresql.org/docs/9.2/static/install-procedure.html

//create table
CREATE TABLE tmp (
    id         integer,
    col1       double,
    col2       double
);

//insert data 
insert into tmp values (1, 2.0,3.0)
insert into tmp values (2, 3.0,4.0)

//execute query
//SGB-All
select  avg(col1) as centerx, avg(col2) as centery
min(col1) as p1x, min(col2) as p1y
max(col1) as p2x, max(col2) as p2y from tmp
group by col1, col2 distanceall within 2
on_overlap any 
metric lone;

select  avg(col1) as centerx, avg(col2) as centery
min(col1) as p1x, min(col2) as p1y
max(col1) as p2x, max(col2) as p2y from tmp
group by col1, col2 distanceall within 2
on_overlap eliminate 
metric lone;

select  avg(col1) as centerx, avg(col2) as centery
min(col1) as p1x, min(col2) as p1y
max(col1) as p2x, max(col2) as p2y from tmp
group by col1, col2 distanceall within 2
on_overlap new_group 
metric lone;             
   			 

//SGB-Any	 
select  avg(col1) as centerx, avg(col2) as centery
min(col1) as p1x, min(col2) as p1y
max(col1) as p2x, max(col2) as p2y from tmp
group by col1, col2 distanceany within 2
metric lone;