PostgreSQL Using multiple columns to calculate numbers for a percentage

Question

I'm very new to SQL and I've been struggling with a pretty basic problem. I've tried to come at this a few different ways but nothing so far has worked.

So in my table touristdata I have 4 columns: transportType, month, year and count. Example:

Month  Year   Transport Type  Count
Nov    1992   Car             100
Nov    1992   Plane           250
Dec    1992   Car             200
Dec    1992   Plane           250
Jan    1993   Car             200
Jan    1993   Plane           200

Except I actually have four different transport types and many more months and years.

I want to calculate the percentage of each transport used over the years. So my desired output would be something along the lines of:

Year   Transport Type  Percentage
1992   Car             37.5% 
1992   Plane           62.5%
1993   Car             50%
1993   Plane           50%

My current code looks like this:

WITH t1 as(
select transport, SUM(ncount) AS transportTotal
from touristdata
GROUP BY transport)
SELECT years, touristdata.transport, ROUND(100.0 *(transportTotal/SUM(ncount)))
FROM touristdata, t1
GROUP BY years;

In this form I get the error:

ERROR:  column "touristdata.transport" must appear in the GROUP BY clause or be used in an aggregate function
LINE 5: SELECT years, touristdata.transport,  ROUND(100.0 *(transpor...

But I know that adding touristdata.transport and transportTotal in to the GROUP BY won't work either. I tried it to make sure and I ended up with 4 entries for each transport type per year.

I didn't have the final 'GROUP BY years' before, and I tried to do it with subqueries but I couldn't figure it out.

If anyone could help me get my head around this I'd really appreciate it!

Answer 1

You can do this using window functions:

SELECT td.years, td.transport, SUM(ncount),
       SUM(ncount) / SUM( 1.0*SUM(ncount) ) OVER (PARTITION BY year) as ratio
FROM touristdata td
GROUP BY td.years, td.transport;