Pairing using associations

Hello,

I’m trying to find pairs in my data.

A pair is two (and only two) products – both were bought by two different customers.

The more customers bought the same two products, the stronger the pair is.

So the rules I’m looking for should be something like that:

X <=> Y & importance of the rule. That’s it.

I thought association algorithm will do the work and I tried to play with the parameters, but I just couldn’t get there.

Is it possible doing it with association algorithm?

(If it’s not possible) what are my options? Writing a plug-in?

You should be able to simply set MAXIMUM_ITEMSET_SIZE to 2 and MINIMUM_SUPPORT to 2. This will give you all of the pairs of items that appear in at least two transactions. Note that this will likely be very large.

If you want every itemset to turn into a rule, you can set MINIMUM_PROBABILITY to some absurdly low number - e.g. 0.000001

Pairing Group

Hello, I have an interesting problem. I have to compare 2 consecutive
records. I'm using Crystal Reports v.10 if that helps. I have a nchar Col4
and a date/time in Col3. I need to compare the date/time of every 2nd of 2
records. The time in the second record of each pair is always a few seconds
,
or a few minutes, greater than in the first record. Col1 and Col2 have
values that are always the same; Col4 varies slightly by the text message.
Any help would be appreciated. Thanks.This would be a self join. For example:
SELECT T1.col1, T1.col2, T1.col3, T1.col4,
MIN(T2.col3)
FROM YourTable AS T1
JOIN YourTable AS T2
ON T1.col3 < T2.col3
GROUP BY T1.col1, T1.col2, T1.col3, T1.col4
David Portas
SQL Server MVP
--|||David,
This query is helpful but I suppose I should have added that I need to CALC
a subtraction of the 1st row's date/timestamp from the 2nd row's
date/timestamp to produce an interval value such as :03 (seconds) etc. Is
this possible? Thanks.
"David Portas" wrote:

> This would be a self join. For example:
> SELECT T1.col1, T1.col2, T1.col3, T1.col4,
> MIN(T2.col3)
> FROM YourTable AS T1
> JOIN YourTable AS T2
> ON T1.col3 < T2.col3
> GROUP BY T1.col1, T1.col2, T1.col3, T1.col4
> --
> David Portas
> SQL Server MVP
> --
>
>|||SELECT T1.col1, T1.col2, T1.col3, T1.col4,
MIN(T2.col3), DATEDIFF(S,T1.col3,MIN(T2.col3))
FROM YourTable AS T1
JOIN YourTable AS T2
ON T1.col3 < T2.col3
GROUP BY T1.col1, T1.col2, T1.col3, T1.col4
David Portas
SQL Server MVP
--|||This was what I needed and it worked great. I marked it a Helpful Post.
Thanks!
"David Portas" wrote:

> SELECT T1.col1, T1.col2, T1.col3, T1.col4,
> MIN(T2.col3), DATEDIFF(S,T1.col3,MIN(T2.col3))
> FROM YourTable AS T1
> JOIN YourTable AS T2
> ON T1.col3 < T2.col3
> GROUP BY T1.col1, T1.col2, T1.col3, T1.col4
> --
> David Portas
> SQL Server MVP
> --
>
>