Market Basket Analysis - Operators to change data layout

ankranock

Hi,

 

I am new to Rapid Miner and I am trying to create a Market Basket Analysis Model using the FP-Growth Operator and the Create Association Operator.

 

I am reading a csv file into Rapid Miner and it looks like the Figure 1 below with attributes going out to att32 and 9,835 observations.  Each row represents a transaction.

Figure 1

Row	att1	att2	att3	att4	att5
1	tropical fruit	yogurt	coffee	?	?
2	whole milk	?	?	?	?
3	pip fruit	yogurt	cream cheese	meat spreads	?
4	other vegetables	whole milk	condensed milk	long life bakery product	?
5	whole milk	butter	yogurt	rice	abrasive cleaner

 

 

I believe the FP-Growth Operator is expecting an example data set like Figure 2 shown below.  Each id corresponds to an item.  The table expands to 167 items and 43,367 rows.

 

Row	id_1.0	id_10.0	id_11.0	id_12.0	id_13.0	id_14.0	Tran_ID
1	true	false	false	false	false	false	1
2	false	false	true	false	false	false	1
3	false	false	false	false	true	false	2
4	false	true	false	false	false	false	2
5	false	false	false	false	false	true	2

 

Are there operators within Rapid Miner that can transform the data from what’s in Figure 1 to something like what’s in Figure 2 that the FP-Growth Operator will like? If I can do it with item names instead of item numbers it would be even better.  I had to transform the layout of the data outside of Rapid Miner to make it work.

 

Thanks for any help or guidance you can provide.

MartinLiebig

Mh,

 

good question! I would try De-Pivot on att.* to get a SQL-ish table for it, and then use Pivot and Numerical to Binominal to get what i what. Not sure if it really works but i think it does.

 

~Martin

ankranock

Martin,

 

I will play with the operators you mentioned and see if I can get them to work.  Thanks for the feedback

 

--Alan--