Question

I have a table with the following columns

source_title, country, language, source_url

I need to generate a query that will give me the following:

country, source_title count, percentage of sources

and

language, source_title count, percentage of sources

basically map the country to all sources and get the count and percentages of this mapping

not the row level data like

SELECT [source_id]
  ,[source_title]
  ,[source_url]
  ,[moreover]
  ,[country]
  ,[lang]
FROM [NewsDatabase].[dbo].[NewsSourcesMatch]
order by country

For example if there are 10 records where country is USA then

country    count(source_title)   % source_title
USA            10                    10/1000 * 100

sorry everyone here is sample data

source_title source_url moreover country lang

Hadeland http://www.hadeland.net Hadeland NORWAY Norwegian

Business Wire http://www.businesswire.com Business Wire UNITED STATES English

Adelaide Now http://www.adelaidenow.com.au Adelaide Now AUSTRALIA English

MSNBC Local http://www.msnbc.msn.com MSNBC Local UNITED STATES English

UDN.com http://forum.udn.com UDN.com TAIWAN Chinese

CBS3 Philadelphia http://cbs3.com CBS3 Philadelphia UNITED STATES English

104.7 Edge Radio http://www.1047edgeradio.com 104.7 Edge Radio UNITED STATES English

so there are four from UNITED STATES so shouldnt the total percentage be 4/7* 100

Answer 1

You can use the OVER clause to span the entire dataset with COUNT to give total number of rows in the same query. Then you have both counts (per country and all rows) to generate the %

Should be something like:

SELECT  [Country]
    ,   [source_title_count] =  COUNT(*)
    ,   [source_total_count]  = COUNT(*) OVER ()
    ,   [source_percent]  = 100.0 * COUNT(*) / COUNT(*) OVER () 
FROM [dbo].[NewsSourcesMatch]
GROUP   BY [Country]

SELECT  [lang]
    ,   [source_title_count] =  COUNT(*)
    ,   [source_total_count]  = COUNT(*) OVER ()
    ,   [source_percent]  = 100.0 * COUNT(*) / COUNT(*) OVER () 
FROM [dbo].[NewsSourcesMatch]
GROUP   BY [lang]

If not, please add sample data and required output.

Or this?

SELECT  [Country]
    ,   COUNT(DISTINCT [source_title)) AS source_title_count
    ,   COUNT(*) source_country_count
    ,   100.0 * COUNT(*) / COUNT(DISTINCT [source_title)) source_country_count
FROM [dbo].[NewsSourcesMatch]
GROUP  BY [Country]

Can’t test this (no SQL on this PC) but based on MSDN OVER clause

SELECT  [Country]
    ,   [source_title_count] =  COUNT(*)
     --attempt 1
    ,   [source_total_count]  = COUNT(*) OVER (Country)
    ,   [source_percent]  = 100.0 * COUNT(*) / COUNT(*) OVER (Country) 
     --attempt 2
    ,   [source_total_count]  = COUNT(*) OVER (PARTITION BY Country)
    ,   [source_percent]  = 100.0 * COUNT(*) / COUNT(*) OVER (PARTITION BY Country) 
FROM [dbo].[NewsSourcesMatch]
GROUP   BY [Country]