根据SQL Server中的条件对行进行计数

Question

I have an example data as follows:

field1  onlineoffline   trxndate
--------------------------------
F1       Offline        2018-04-03
F1       Online         2018-04-04
F1       Online         2018-04-05
F1       Offline        2018-04-06
F1       Offline        2018-04-07
F1       Offline        2018-04-08
F1       Offline        2018-04-09
F1       Online         2018-04-10
F1       Offline        2018-04-11

I need to get count of records which are online after 1st online based on the trxndate(which is 2 in this example), also I want to get count of all the records after 1st online record based on the trxndate(7 in this example) in a single query.

I tried using the row_number() but didn't quite get to the point. Is there a way to get this.

Answer 1

Is this what you want?

select count(*) as num_records,
       sum(case when onlineoffline = 'online' then 1 else 0 end) as num_online_record
from (select t.*,
             min(case when onlineoffline = 'online' then trxndate end) over () as min_online_td
      from t
     ) t
where trxndate > min_online_td;

The subquery calculates the date for the first online record. The rest of the query just does the counts that you want.

Answer 2

select  -- count only Online rows
        online_count = sum(case when onlineoffline = 'Online' then 1 else 0 end),
        -- count all rows
        all_count = count(*)
from    tbl
where   trxdate >   
        -- find the first trxdate that is online
        (select top 1 trxdate from tbl where onlineoffline = 'Online' order by trxdate)

Answer 3

Determine the first online date, then do 2 counts.

;WITH MinimumOnlineDate AS
(
    SELECT
        MinDate = MIN(T.trxndate)
    FROM
        YourTable AS T
    WHERE
        T.onlineoffline = 'Online'
)
SELECT
    TotalAfterFirstOnline = COUNT(1),
    TotalOnlineAfterFirstOnline = COUNT(CASE WHEN T.onlineoffline = 'Online' THEN 1 END)
FROM
    YourTable AS T
    INNER JOIN MinimumOnlineDate AS M ON T.trxndate > M.MinDate

Answer 4

other option is to do with correlation subquery

select count(*) as online_counts, max(t1.counts) as no_of_onlines
from table t
join (
    select distinct min(trxndate) over() trxndate, count(*) over (order by field1) Counts 
    from table t
    where onlineoffline = 'Online' and trxndate <> (
          select min(trxndate) from table where onlineoffline = 'Online' and field1 = t.field1) 
)t1 on t.trxndate >= t1.trxndate

Answer 5

select
    sum(case when onlineoffline = 'Online' then 1 else 0 end) online,
    count(*) all 
from    table
where   trxdate > (select top 1 trxdate from table where onlineoffline = 'Online' order by trxdate)