The evaluation randomly splits the dataset into training and test set. You see different results for different splits.
--sebastian On 27.11.2011 15:30, bish maten wrote: > the Evaluation Result is changing each time it is run with mvn exec command. > Why is the result changing each time it run? Evaluation result varies from > 1.0, 0.66, 1.2, ... > > > INFO: Unable to recommend in 2 cases > Nov 27, 2011 6:22:43 AM org.slf4j.impl.JCLLoggerAdapter info > INFO: Evaluation result: 1.0 > Nov 27, 2011 6:22:43 AM org.slf4j.impl.JCLLoggerAdapter info > INFO: 1.0 > > another run gives following > > INFO: Unable to recommend in 8 cases > Nov 27, 2011 6:25:08 AM org.slf4j.impl.JCLLoggerAdapter info > INFO: Evaluation result: 0.6666666666666666 > Nov 27, 2011 6:25:08 AM org.slf4j.impl.JCLLoggerAdapter info > INFO: 0.6666666666666666 > > > >> the output from GroupLensRecommenderEvaluatorRunner gives following output >> (given a input rating file) >> >> Evaluation Result 1.09 >> and Unable to recommend in two cases >> >
