Analysing which products are sold together

[siwe]

Hi,

i’m a absolute beginner at KNIME and want to analyse, which products are sold togehter. The aim is to recommend a customer a product, because other customers bought the same product combination.

I have a table with product numbers and date, which shows all transactions of customers over a certain amount of time. It looks like this:

Customer Date Product 1 Product 2 Product 3 … Product 32
2568 01/04/20 BC-14569 BA-5698 DFK-236
5897 15/03/20 TF-25984 1257-DFR
1269 17/03/20 C1254 25-Z1267 457-GJHG 415ABG

You can read it like: Customer 2568 bought product “BC-14569”, product “BA-5698” and product “DFK-236” at that day.

Now i want to create a matrix with all products in the columns and the rows. It should show how often two products have been bought at the same time, so I can see which two products are most likely to be bought togehter. I tried it in excel with a matrix function. It worked, but it was too much data to be calculated.

I hope someone has an idea to solve this. Thank you!

Kind regards

[tommy]

Hi @siwe and welcome to the KNIME world,

please have a look at https://hub.knime.com/tommy/spaces/Public/latest/Cross-selling%20Matrix.
This should give you the product matrix with the cross-selling potential.
First step should be to have the purchase (customer, date) and one product unique in one row. Afterwards you can build pairs of product and count the purchases.

Hope that helps as a start,
Greetz, Tommy