Predictive model

Hi guys,

Is there a possibility to build a prediction model with training and test data and corresponding algorithms. the data I have should be labelled according to my understanding or?

Please help me :slight_smile:


Hi @anon33357744,

I had a quick look at your workflow. It’s of course possible to train a model, yet can you be more clear about what it should do? As of now, I just see that you are doing some filtering after a join… Would you mind explaining in a few words what your approach is intended to do? :wink:


1 Like

HI @Alec,

first of all thank you very much that you want to help me :slight_smile:
I have an Excel sheet with different information (columns) about transactions. I would like to be able to say in the future whether a new transaction is a completely new constellation or not.


Hi @anon33357744,

so if I may rephrase your intention: you want to train a model which detects whether a transaction fits into the pattern of previous transactions or whether a transaction is rather unusual, right?
In that case, I’d like to refer you to this page, where several data mining techniques to achieve this goal are listed. I personally would try a first shot with an isolation forest (I believe you’ll find it in the Weka nodes), but there are plentiful techniques all having their pros and cons.


This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.