Which variables should i use/create for a prediction model?

Hello @cdaponte

Did you try rapidminer automodel? It will suggest models and also attributes useful in making predictions. You can also understand which attributes are important in making predictions.

Give it a try.

Thanks! Yes i already try it but, wow it´s very complex 

Do you mean the operator connections in the model are very complex (or) to understand what is happening inside the process is complex?

It´s difficult for me to understand what is happening inside the process. 

I don't want to push you, but if you are interested you can take a look at these videos that might help you understand what is happening in the auto model. If you have more questions on the auto model, you can ask us here and we are happy to answer them.

https://academy.rapidminer.com/learn/video/auto-model-classification
https://rapidminer.com/resource/automated-machine-learning/
https://www.youtube.com/watch?v=Ol0ZXN-GFTo

Hope this helps.

Hi:  After following varunm1's suggestions above, it might also be very helpful to share a summary of your RapidMiner process outputs with people within your organization who have familiarity with the business issues - i.e. people who have been around awhile and have a feel for factors that may play a role in determining whether or not a given loan has a risk of defaulting.  Discussing model outputs with experts also is a way of building trust in the models so that the organization will be more likely to deploy them and integrate predictive model deliverables into other data flows within the organization.  Best wishes, Michael Martin 

Hi @cdaponte

Regarding your dataset - as I understand it contains some historical data about Santander bank borrowers, right?
Which column represents a target variable (the one you are trying to predict, whether the debtor has paid back)? Or maybe it should be derived somehow from other attributes? It's not easy to understand as column names formed from Spanish language not known to me.