Solving MAX(COUNT()) problem 2 – optimizations

In previous post I’ve tried to present my solution for solving max(count()) problem.  The solution was slightly suboptimal and I’ve needed to speed it up a bit, because I’m using it now in some statistical calculations and every millisecond is important. This is original solution. CREATE VIEW data_view_source AS SELECT DISTINCT ON (dp_id) dp_id, ds_id […]

SQL timeline and statistical computations

This is simple way how-to generate time table (list of consequencing timestamps, or timeline) in PostgreSQL. Nothing spectacular, but might help you ,when trying to do some time based statistical selects (can feel like OLAP :-)). Code: CREATE OR REPLACE FUNCTION gen_time_list( tim IN TIMESTAMP, tim_stop IN TIMESTAMP, step IN INT) RETURNS TABLE(ts1 TIMESTAMP, ts2 […]

Solving MAX(COUNT()) problem

I’ve been solving problem of doing grouped MAX(COUNT()) in PostgreSQL, and because I’ve not found anything really usable out there (doing correlated sub-queries is definitely not good idea for thousands of records) I’ve had to find my own solution. Situation plan My configuration is a bit complicated to explain this, so i’ll try do it […]