There is no such thing as multiclass regression. These metrics are for
classification problems and don't have meaning for regression.
On Tue, Dec 6, 2016 at 7:55 PM Md. Rezaul Karim <
rezaul.ka...@insight-centre.org> wrote:
> Hi Sean,
>
> According to Spark documentation, precision, recall, F1, true positive
> rate, false positive rate etc. can be calculated using the MultiMetrics
> evaluator for the multiclass classifiers also. For example in *Random
> Forest *based classifier or regressor:
>
> // Get evaluation metrics.
> MulticlassMetrics metrics = new
> MulticlassMetrics(predictionAndLabels.rdd());
> //System.out.println(metrics.confusionMatrix());
> // System.out.println(metrics.confusionMatrix());
> double precision = metrics.precision(metrics.labels()[0]);
> double recall = metrics.recall(metrics.labels()[0]);
> double f_measure = metrics.fMeasure();
> double query_label = 2001; // it's a label or class for prediction
> double TP = metrics.truePositiveRate(query_label);
> double FP = metrics.falsePositiveRate(query_label);
> double WTP = metrics.weightedTruePositiveRate();
> double WFP = metrics.weightedFalsePositiveRate();
>
> Where the related performance measure statistics is calculated and stored
> in '*predictionAndLabels*' RDD as follows:
> JavaRDD<Tuple2<Object, Object>> predictionAndLabels = testData.map(
> new Function<LabeledPoint, Tuple2<Object, Object>>() {
> public Tuple2<Object, Object> call(LabeledPoint p) {
> Double prediction = model.predict(p.features());
> return new Tuple2<Object, Object>(prediction, p.label());
> }
> }
> );
> And *'model'* is a Random Forest model instance trained with multiclass
> regression or classification dataset.
>
> The current implementation of Logistic Regression supports only the binary
> classification. But, Linear Regression supports/works on the dataset having
> multiclass.
>
> I was wondering if it's possible to compute the similar metrics using the
> Linear Regression based model for multiclass or binary class dataset.
>
>
>
> Regards,
> _________________________________
> *Md. Rezaul Karim* BSc, MSc
> PhD Researcher, INSIGHT Centre for Data Analytics
> National University of Ireland, Galway
> IDA Business Park, Dangan, Galway, Ireland
> Web: http://www.reza-analytics.eu/index.html
> <http://139.59.184.114/index.html>
>
> On 6 December 2016 at 11:37, Sean Owen <so...@cloudera.com> wrote:
>
> Precision, recall and F1 are metrics for binary classifiers, not
> regression models. Can you clarify what you intend to do?
>
> On Tue, Dec 6, 2016, 19:14 Md. Rezaul Karim <
> rezaul.ka...@insight-centre.org> wrote:
>
> Hi Folks,
>
> I have the following code snippet in Java that can calculate the precision
> in Linear Regressor based model.
>
> Dataset<Row> predictions = model.transform(testData);
> long count = 0;
> for (Row r : predictions.select("features", "label",
> "prediction").collectAsList()) {
> count++;
> }
> System.out.println("precision: " + (double) (count * 100) /
> predictions.count());
>
> Now, I would like to compute other evaluation metrics like *Recall *and
> *F1-score
> *etc. How could I do that?
>
>
>
> Regards,
> _________________________________
> *Md. Rezaul Karim* BSc, MSc
> PhD Researcher, INSIGHT Centre for Data Analytics
> National University of Ireland, Galway
> IDA Business Park, Dangan, Galway, Ireland
> Web: http://www.reza-analytics.eu/index.html
> <http://139.59.184.114/index.html>
>
>
>
