"Issue in getting the predicted label from a Java application"

spg

Hi,


I created a classification process using Rapidminer, now I'm trying to get the predicted label from this process in my JAVA application, for that I am using the following code:

RapidMiner.setExecutionMode(RapidMiner.ExecutionMode.COMMAND_LINE);
RapidMiner.init();

com.rapidminer.Process process = new com.rapidminer.Process(new File("readModel.rmp"));

IOContainer results = process.run();
results.asList();

ExampleSet resultSet2 = results.get(ExampleSet.class);

for (Example example : resultSet2) {

Attribute predictedLabel = example.getAttributes().getPredictedLabel();
predictionLabel = Double.parseDouble(example.getNominalValue(predictedLabel));

System.out.println(predictionLabel);
}

The problem is that the predicted label that I receive in my application is different from the predicted label from the rapidminer.

I am using the same files and data in both sides.

Thanks for any help.

Marco_Boeck

Hi,

please be more specific. What do you mean by different? 0.349745 instead of 0.35? Or 0.1 instead of 0.7? Also please post the process XML here.

Regards,
Marco

spg

Thanks for the reply,

My classification process classifies new data into one of four possible classes (1.0, 2.0, 3.0 or 4.0).
I'm trying get the prediction label in my JAVA application.
The problem is that sometimes for example I get the prediction label 1.0, but if I run the calssification process with the same new data (test data) in Rapidminer I get the prediction label 3.0.
I don't know why this happens, I think the results should be the same.

< parameter key="logverbosity" value="init"/>
< parameter key="random_seed" value="2001"/>
< parameter key="send_mail" value="never"/>
< parameter key="notification_email" value=""/>
< parameter key="process_duration_for_mail" value="30"/>
< parameter key="encoding" value="SYSTEM"/>


< parameter key="excel_file" value="C:\Test Data.xls"/>
< parameter key="sheet_number" value="1"/>
< parameter key="imported_cell_range" value="A1:L3"/>
< parameter key="encoding" value="SYSTEM"/>
< parameter key="first_row_as_names" value="false"/>

< parameter key="0" value="Name"/>

< parameter key="date_format" value=""/>
< parameter key="time_zone" value="SYSTEM"/>
< parameter key="locale" value="English (United States)"/>
<列出关键= " data_set_meta_data_information " >
< parameter key="0" value="ID.true.integer.id"/>
< parameter key="1" value="Projecto.true.binominal.attribute"/>
< parameter key="2" value="NUM_T.true.integer.attribute"/>
< parameter key="3" value="NUM_M.true.integer.attribute"/>
< parameter key="4" value="META.true.binominal.attribute"/>
< parameter key="5" value="INSTANT.true.integer.attribute"/>
< parameter key="11" value="LABEL.true.integer.label"/>

< parameter key="read_not_matching_values_as_missings" value="true"/>
< parameter key="datamanagement" value="double_array"/>


< parameter key="model_file" value="C:\model.mod"/>



< parameter key="create_view" value="false"/>

Thanks for any help

Marco_Boeck

Hi,

what is the actual model you are using? Your process hides that quite well
Also it would be quite helpful if you provided a sample of the data with which I could replicate this behavior.

Regards,
Marco

spg

I'm reading the model from another file, here is the XML:

http://pastebin.com/q4Fq021j

Data sample:

ID Project NUM_T NUM_M INSTANT C_EST LABEL

2196 A 13 3 81 132 1
2197 B 14 3 51 136 2
2198 C 15 3 94 145 2
2199 D 16 3 48 152 2
2200 E 17 3 12 136 3
2201 F 18 3 100 151 2
2202 G 19 3 22 162 3
2203 H 20 3 66 177 4
2204 I 21 3 130 184 2
2205 J 22 3 29 199 3
2206 M 23 3 90 213 2
2207 N 6 4 9 21 4
2208 O 8 5 55 129 2
2209 P 9 3 2 95 1
2210 Q 6 4 2 17 2


The model is using just the vars C_EST, NUM_M and INSTANT.

Thanks for any help

Marco_Boeck

Hi,

the following code produces the exact same result with your model and your sample data regardless whether I use the GUI or a custom Java program to execute it.

RapidMiner.setExecutionMode(ExecutionMode.COMMAND_LINE);
RapidMiner.init();

// loads the process from the repository (if you do not have one, see alternative below)
RepositoryLocation pLoc = new RepositoryLocation("//Local Repository/testProcess");
ProcessEntry pEntry = (ProcessEntry) pLoc.locateEntry();
String processXML = pEntry.retrieveXML();
com.rapidminer.Process myProcess = new com.rapidminer.Process(processXML);
myProcess.setProcessLocation(new RepositoryProcessLocation(pLoc));
IOContainer ioResult = myProcess.run();

// use the result(s) as needed, for example if your process just returns one ExampleSet, use
// this:
if (ioResult.getElementAt(0) instanceof ExampleSet) {
ExampleSet resultSet = (ExampleSet) ioResult.getElementAt(0);
RepositoryManager.getInstance(null).getRepository("Local Repository")
.createIOObjectEntry("testResult", resultSet, null, null);
System.out.println("Stored");
}

Regards,
Marco