The Data Mining Forum 
IMPORTANT: This is the old Data Mining forum.
I keep it online so that you can read the old messages.
Please post your new messages in the new forum: https://forum2.philippe-fournier-viger.com/index.php
Analysis Services and creating a market basket analysis
Posted by:
Azake
Date: July 09, 2013 06:22AM
Hello,
actually I am trying to make a market basket analysis with SQL Server 2012 and its tools "Server Data Tools" and "SQL server Management studio" (Visual Studio)
As basis I have a table with different types of products including customer attributes.
I want to use the association mining rules.
Here you can see the first 2 Datasets
TransactionID Frozenfoods Alcohol Snacks GENDER Age MARITAL CHILDREN WORKING
1 1 0 1 Female 18 to 30 Widowed No Yes
2 1 0 0 Female 18 to 30 Separated No Yes
You can see there will be only differentiated between "bought" or "not bought" (0 or 1) within the products.
I am not sure if I have to create two tables ( Products / customer ).
Moreover I am not sure about setting up the right inputs and predictions.
My goal is to have a real market basket analysis but now I got some confusing results. I guess somewhere I didn't set it up right.
I already made this with the IBM SPSS Modeler (A Prior and Carma) and this was quite "easy".
I hope you can help me because I already spend days of just finding help. I would say I have problems with these binary itemsets. 0 and 1.
Thanks
Edited 1 time(s). Last edit at 07/09/2013 07:34AM by Azake.
actually I am trying to make a market basket analysis with SQL Server 2012 and its tools "Server Data Tools" and "SQL server Management studio" (Visual Studio)
As basis I have a table with different types of products including customer attributes.
I want to use the association mining rules.
Here you can see the first 2 Datasets
TransactionID Frozenfoods Alcohol Snacks GENDER Age MARITAL CHILDREN WORKING
1 1 0 1 Female 18 to 30 Widowed No Yes
2 1 0 0 Female 18 to 30 Separated No Yes
You can see there will be only differentiated between "bought" or "not bought" (0 or 1) within the products.
I am not sure if I have to create two tables ( Products / customer ).
Moreover I am not sure about setting up the right inputs and predictions.
My goal is to have a real market basket analysis but now I got some confusing results. I guess somewhere I didn't set it up right.
I already made this with the IBM SPSS Modeler (A Prior and Carma) and this was quite "easy".
I hope you can help me because I already spend days of just finding help. I would say I have problems with these binary itemsets. 0 and 1.
Thanks
Edited 1 time(s). Last edit at 07/09/2013 07:34AM by Azake.
Re: Analysis Services and creating a market basket analysis
Posted by:
webmasterphilfv
Date: July 09, 2013 02:39PM
Hi,
An association rule mining algorithm will find association between groups of items occuring in a transaction transaction database, where a transaction is a set of items. Each item can be either present or absent (0 or 1).
In your table, the products are represented as binary values which is good.
For the other columns, the data is not binary. If you want to include these columns in the associations that you want to discover, then you would need to convert them to binary or to use an association rule mining algorithm that accept non binary attributes (there probably exists some). Or you could just ignore them. It depends on what you want to do with the patterns.
Hope this helps a little bit,
Philippe
An association rule mining algorithm will find association between groups of items occuring in a transaction transaction database, where a transaction is a set of items. Each item can be either present or absent (0 or 1).
In your table, the products are represented as binary values which is good.
For the other columns, the data is not binary. If you want to include these columns in the associations that you want to discover, then you would need to convert them to binary or to use an association rule mining algorithm that accept non binary attributes (there probably exists some). Or you could just ignore them. It depends on what you want to do with the patterns.
Hope this helps a little bit,
Philippe
Re: Analysis Services and creating a market basket analysis
Posted by:
Azake
Date: July 10, 2013 01:35AM
Hey Philippe,
thanks a lot.
In my imagination I thought I could show/see a dependency with a group of customer who buy A and when they buy A they buy with a confidence of xx% Product B.
In the SPSS Modeler from IBM I could see dependencies like If "tinned goods = 1" Then Bakery Goods = 1 And Snacks = 1 for example.
This is good but I thought I can see it more like a decision tree algorithm. Maybe I have the wrong imagination of this algorithm.
Furthermore I am facing problems with the binaries. All the goods/products have 0 or 1. So far so good but I think I set the contenttypes wrong. Sometimes my results are like "Tinned goods = 0 - 1" which is wrong I guess. I set it up as discretized.
MS Analysis Services is kinda confusing and complicated for a newbie to be honest.
I hope I could get further help.
Thanks for the support so far.
Max
thanks a lot.
In my imagination I thought I could show/see a dependency with a group of customer who buy A and when they buy A they buy with a confidence of xx% Product B.
In the SPSS Modeler from IBM I could see dependencies like If "tinned goods = 1" Then Bakery Goods = 1 And Snacks = 1 for example.
This is good but I thought I can see it more like a decision tree algorithm. Maybe I have the wrong imagination of this algorithm.
Furthermore I am facing problems with the binaries. All the goods/products have 0 or 1. So far so good but I think I set the contenttypes wrong. Sometimes my results are like "Tinned goods = 0 - 1" which is wrong I guess. I set it up as discretized.
MS Analysis Services is kinda confusing and complicated for a newbie to be honest.
I hope I could get further help.
Thanks for the support so far.
Max