Your reasoning is correct; you need probabilities (or at least some score) out of the model and not just a 0/1 label in order for a ROC / PR curve to have meaning.
But you just need to call clearThreshold() on the model to make it return a probability. On Tue, Nov 24, 2015 at 5:19 PM, jmvllt <mouvilliat.j...@gmail.com> wrote: > Hi guys, > > This may be a stupid question. But I m facing an issue here. > > I found the class BinaryClassificationMetrics and I wanted to compute the > aucROC or aucPR of my model. > The thing is that the predict method of a LogisticRegressionModel only > returns the predicted class, and not the probability of belonging to the > positive class. So I will get: > > val metrics = new BinaryClassificationMetrics(predictionAndLabels) > val aucROC = metrics.areaUnderROC > > with predictionAndLabels as a RDD[(predictedClass,label)]. > > Here, because the predicted class will always be 0 or 1, there is no way to > vary the threshold to get the aucROC, right ???? Or am I totally wrong ? > > So, is it relevant to use BinaryClassificationMetrics.areUnderROC with > MLlib's classification models which in many cases only return the predicted > class and not the probability ? > > Nevertheless, an easy solution for LogisticRegression would be to create my > own method who takes the weights' vector of the model as a parameter and > computes a predictionAndLabels with the real belonging probabilities. But is > this the only solution ???? > > Thanks in advance. > Regards, > Jean. > > > > > -- > View this message in context: > http://apache-spark-user-list.1001560.n3.nabble.com/Is-it-relevant-to-use-BinaryClassificationMetrics-aucROC-aucPR-with-LogisticRegressionModel-tp25465.html > Sent from the Apache Spark User List mailing list archive at Nabble.com. > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@spark.apache.org > For additional commands, e-mail: user-h...@spark.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
