RProp MLP

Hi, im new using KNIME and Im trying to use RProp MLP, but i'm getting this warn:

WARN      StringToNumberNodeModel            Values in 2135 cells could not be parsed, first error: 'Masculino' (RowKey: Row0, Position: 0)
WARN      String To Number                   Problems occurred, see Console messages.

What can I do?  Thanks.

Hi,

I think you are trying to convert strings to doubles as input for MLP learner.

I attach an example workflow. I hope this helps.

Here is the workflow :))

While using the RProp MLP leaner the first warn was:

WARN      RProp MLP Learner                  Only double columns for input

So the solution that i found in the forum was to add the String to Number node in order to convert the data into double. And it worked. Now there's another warn:

WARN      StringToNumberNodeModel            Values in 2135 cells could not be parsed, first error: 'Masculino' (RowKey: Row0, Position: 0)

I guess the other values (nominal) are not being read. 

You have to convert nominal values to numbers using a similar workflow like the one I attached.

String to Number node works if you have numbers in columns which are coded as strings.

If possible, attach an example of your workflow here.

Here it is :) 

Here is an example that might help you.

You forget to include your data to workflow, there fore I could not run the flow, but you may.

Here is the data. I'm trying to run it but it wont. 

Im sorry, i think im doing something wrong.. :(

Hope, this helps.

Thanks for the example boraster.  For the sake of completeness, I find it very nice to use the One to Many (PMML) node to prepare nominal values for use in neural networks.  You can additionally use the Normalizer (PMML) node and feed both of these PMML fragments into the MLP Learner to have these data manipulation steps embedded in the model.  This is very convenient as if you use a predictor which supports these fragments, such as the JPMML Classifier, as you can just feed in the data to score directly into the predictor and have the numberificatioin and normalization steps happen automatically. 

Linked below is a youtube video which covers some new features in this area in KNIME 2.11:

https://www.youtube.com/watch?v=yQP0NImUCes

 

Hi Aaron,

can you please give me your e-mail address? I would like to share something with you.

regards.

 

bora