Amé,
being fresh from a training course on intelligent miner, I don't think
it will work quite like this. The associations model accepts just two
columns at once.(assuming you're using the association model in the
miner)
So assume you had a list of transactions, with different SKU ids
against each one. The association model will look for combinations of
particular SKUs (eg Apple with Pear) within the same transaction
(whether that's defined as a particular shopping basket or a particular
person) but it doesn't look for quantity. That is, it doesn't matter
if you've purchased one pear or twenty pears; either way, you're a
pear-customer.
Obviously this doesn't give very useful associations, so you need to
set up taxonomies on top of the SKUs. This would allow you to group
together all fruit products, say, and then see if there's an
association between fruit and other products.
I suppose you could build a taxonomy on quantity and make the SKU a
combination of SKU + quantity - that might help. But somebody with
more real-world experience than me of this will probably have another
take on this.
Regards
JCSJF
Amé wrote:
Hi, I am new to DB2 - Intelligent miner and I would like to perform
some mining on data. My problem is that I have transactions of the
following form
Transaction ClientID Unit Article
1 105 1 Apple
1 105 2 Pear
2 482 1 Carrot bag
3 342 6 Orange
and would like to take into account that a product was bought two or
three times... (ex.: client #105 bought 2 pear and 1 Apple..)
Is it handled automatically by Intelligent miner ? What are the correct
setting ?
Thanks a lot!
Amélie Ouellet
Working on :
DB2 7.1
IMD 7.1
Windows XP