Apriori doesent work

Hello,

I have the following problem:

A table in which sold items are listed per row. Each item has an order number. If several items were purchased in one order, they are then displayed twice.

Order number 1 cell phone
Order number 1 charging cable
Order number 1 cell phone cover

Now I want to try to evaluate with the Apriori. Unfortunately, it always tells me that it cannot process strings. Even if I group the order numbers into a list

Order number 1 cell phone cover, cell phone, charging cable

it does not work at all.

Translated with DeepL.com (free version)

See if this helps.

Sadly not, because if i use “Group By” then it converts me all items as a string in a list. When i try to use the “Cell Splitter” it says "No column in spec compatible to “StringValue”.

I use “GroupBy” and select the order number at first and after that i say that every item should be put in a list, because there is no feature that lets me put it in a string.

Someone said “I believe that this is the problem. I think that the data should be formatted like this:” but didint show how i can get there. The biggest problem ist that my table looks like following:

I first need to group everything by the order number, because in the following picture you see, that there are multiple order numbers with different items:

Have you tried the “Split Collection Column” node? I haven’t used it, but it might be worth a try.

2 Likes

This runs with the “Split Collection Column” node. I have absolutely 0 experience setting the Apriori node parameters. Maybe someone with expertise can weigh in. I added an Association Rule Learner node which works directly with lists. Its underlying basis is the Apriori algorithm.

1 Like

It works! Thanks!

  1. “Group by” and make a List of Strings! It doesent work with numbers.

  2. “Split Collection Column” to Split the Lists.

  3. “Column Filter” So i could filter out the order number.

  4. “Missing Value” and make every missing value nothing.

  5. Then i renamed the columns so apriori gives me better information.

  6. U need to play around with the options in apriori, thats mine:

Now i just need to work out some significant numbers… A lift of 3400+ isnt really usable…

If someone has any ideas how to do that, youre welcome to comment!

2 Likes

Glad I could help. Good luck sorting out the parameters.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.