Re: simplest of code

Thanks Peter for the pointers.................

******************************************************************

private class WekaApp {

void doIt() throws Exception {
WekaApp wekaApp= new WekaApp();
       try{
               currentInstance();
           }
           catch (Exception e)
           {
               System.out.println("ERROR: "+ e);
               e.printStackTrace(myConsole.getOut()); // show the stack 
trace
           }

} //end of doIt()

void currentInstance() throws Exception {
try {
// [A] load model and header of data
// deserialize model

Classifier cls = (Classifier) 
weka.core.SerializationHelper.readAll("/somewhere/j48model");

// [B] create an Instances object on the fly and save it to an ARFF file

data = new Instances("MyRelation", atts, 0);

// create the current instance
      values = new double[data.numAttributes()];
// - numeric
      values[0] = OSV;
setDataset(currentInstance);

// [C] classify nominal class

// classifyInstance() just returns the index of the predicted label (the 
one with the highest probability) as a double
prediction = classifyInstance(currentInstance);
// myConsole.getOut().println("prediction: "+ prediction);
}
catch (Exception e)
{
     e.printStackTrace(myConsole.getOut()); // show the stack trace
     myConsole.getOut().println("Weka exception occurred: "+ e);
}

} // end of currentIcurrentstance

} // end of class WekaApp

******************************************************************

Have errors with 'currentInstance' and 'vals'

Bob M

On 11/8/2021 11:51 AM, Peter Reutemann wrote:
>> I am viewing one minute data from the forex market in real time
>>
>> At certain times I wish to
>>
>> 1) recall a saved model
> Use the weka.core.SerializationHelper helper class to load model and
> header of data it was trained with ("readAll" method):
> https://waikato.github.io/weka-wiki/serialization/
>
>> 2) present it with one new instance which comprises just one real variable
> The following article describes how to create an Instances object on
> the fly and saving it to an ARFF file. Since you are loading the
> header (ie empty Instances object) with your model, you can skip
> straight to creating Instance objects ("first instance"). You don't
> need to add the Instance to the Instances (ie header), simply
> reference the header information via "setDataset(Instances)":
> https://waikato.github.io/weka-wiki/formats_and_processing/creating_arff_file/
>
>> 3) have the model do a prediction based on this new instance
> Not sure whether your class is numeric or nominal, so you can either
> use the model's:
> - classifyInstance method to obtain regression value or class label index
> - distributionForInstance method to obtain class distribution
>
> Cheers, Peter
_______________________________________________
Wekalist mailing list -- [email protected]
Send posts to [email protected]
To unsubscribe send an email to [email protected]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html