Home/ Questions/Q 4172346
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-21T00:33:08+00:00

Suppose I have a table that logs incoming users where each user has an

  • 0

Suppose I have a table that logs incoming users where each user has an IP address (ipaddr).

What’s the best way to select all users that have never come to the site before? (so that particular IPADDR value only exists in the table once ), however I only want to know about the new visitors that came in the last 6 hours.

I basically want to do something like this in SQL:

SELECT * from visitors GROUP BY ipaddr WHERE COUNT(ipaddr) = 1 and date > '2011-03-31 00:59:11'

HOWEVER, the DATE condition should only apply to the results, and not for checking if the visitors is new or not.

UPDATE:

There is an SID field that accounts for user browsing sessions.

Here’s the relevant table schema: 

CREATE TABLE `visitors` (
  `date` timestamp NOT NULL default CURRENT_TIMESTAMP,
  `sid` bigint(12) unsigned NOT NULL,
  `ipaddr` int(8) NOT NULL,
)

Some example data:

INSERT INTO `visitors` (`date`,`sid`, `ipaddr`)
VALUES
    ('2011-03-31 06:25:48', 299521885457, -1454342140);


INSERT INTO `visitors` (`date`,`sid`, `ipaddr`)
VALUES
    ('2011-03-31 06:26:37', 299521885457, -1454342140);


INSERT INTO `visitors` (`date`,`sid`, `ipaddr`)
VALUES
    ('2010-01-01 15:23:44', 694387538590, -1454342140);

This visitor has two rows for his current session happening in realtime, each row is for each page he has visited (only relevant schema shown). The last example row shown is a visit from 2010, which means this ip address has 2 different SIDs belonging to it, therefore is not a new visitors.

The query’s result should have none of the rows listed above, seeing as how this visitor has two sessions in the database. If the last row is removed (with sid 694387538590), the visitor should become a new visitor and appear in the query.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    the "WHERE" for GROUP BY is HAVING:

    SELECT ipaddr from visitors
GROUP BY ipaddr
HAVING COUNT(ipaddr) = 1 AND MIN(date) > '2011-03-31 00:59:11'

    UPDATE

    SELECT ipaddr, max(sid) sid
  FROM visitors
 GROUP BY ipaddr
HAVING     COUNT(DISTINCT sid) = 1
       AND MIN(date) > '2011-03-31 00:59:11'

    Explanation:

    SELECT date, sid, ipaddr FROM visitors

date                sid        ipaddr 
------------------------------------------
2011-03-31 06:25:48 299525457  -1454342140 
2011-03-31 06:26:37 299525457  -1454342140 
2010-01-01 15:23:44 694388590  -1454342140 
2011-03-31 11:23:44 111111111  -1234444811 
2011-03-31 12:23:44 111111111  -1234444811

SELECT ipaddr FROM visitors GROUP BY ipaddr

ipaddr
-----------
-1454342140 
-1234444811 

--- group for ip -1454342140 ---

2011-03-31 06:25:48 299525457  -1454342140 
2011-03-31 06:26:37 299525457  -1454342140 
2010-01-01 15:23:44 694388590  -1454342140

COUNT(DISTINCT sid) = COUNT(299525457, 694388590) = 2
--> there is more than 1 session for this ip: not good!!!

  ==> group discarded

--- group for ip -1234444811 ---

2011-03-31 11:23:44 111111111  -1234444811 
2011-03-31 12:23:44 111111111  -1234444811

COUNT(DISTINCT sid) = COUNT(111111111) = 1 --> OK
(here COUNT(sid) = count(111111111, 111111111) = 2
 --> despite it is the same sid, the count is 2, that is why using DISTINCT)

MIN(date) = '2011-03-31 11:23:44' > '2011-03-31 00:59:11' --> OK

  ==> group accepted

    Authorized columns in the SELECT are:

    • columns used in the GROUP BY clause
    • agregates of the other columns

    ipaddr was used in GROUP BY but not sid. To have also sid I used MAX but remember that it will be applied only to the group of rows for current ipaddr and because of the conditions in the query there is 1 unique sid but repeated so the result will be that sid

    • 0