Options

# Multiple Label Prediction

Member Posts: 1 Contributor I
edited November 2018 in Help
First let me apologize, I'm new to Rapid Miner and I'm sure what I'm about to ask is a simple question...

Basically what I'm trying to do is create and test various modeling methods for multi-attribute data.  i.e. I have imported an Excel spreadsheet with six inputs and three outputs (labels?), and I would like to look at the predictions made by various models for these three outputs, based on the training data.  All three outputs are a normalized distribution.  The problem I'm encountering is that I can't seem to assign multiple labels to my data set.  Any suggestions/examples on how to do this?

Here's an example of the data, the last three being the numbers I'm trying to predict.  All inputs are are "real" values ranging from -1 to 1, and all outputs are real values ranging from 0 to 1...

1,0.01480111,-0.719155878,0.2,-0.4,-0.5,0,1,0

-M

• Options
Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
Hi,

this is the wrong place. Please post questions like this in the "Data Mining / ETL / BI Processes" board of this forum. I have this topic moved there for you.

The solution is quite simple: Just name those three columns with a number from 1 to 3 in it, for example "label1", "label2", and "label3". You can then iterate through your process and setting the role for each of the labels on the fly with the "Set Role"-operator.

Cheers,
Ingo
• Options
Member Posts: 1 Contributor I

What if the outputs are dependent on each other

My Scenario

The model should take a set of inputs as a single vector and give output as a single vector.  This single vector of three outputs then needs to be fed to the next stage. So looping might not

Theoretically K-NN should be able to able to predict vectors.

So is there a way to makr multiple labels at the same time instead of looping through the process three time. Interdependencies could be lost