CODEWITHSUNDEEP
pythonapache-sparkpysparkprediction
AlxRd
AlxRd

Reputation: 285

How to obtain probabilities from Pyspark One-vs-Rest multiclass classifier

It does not appear that Pyspark Onv-vs-Rest classifier provides probabilities. Is there a way to do this?

I am appending code below. I am adding the standard multiclass classifier for comparison.

from pyspark.ml.classification import LogisticRegression, OneVsRest
from pyspark.ml.evaluation import MulticlassClassificationEvaluator

# load data file.
inputData = spark.read.format("libsvm") \
    .load("/data/mllib/sample_multiclass_classification_data.txt")

(train, test) = inputData.randomSplit([0.8, 0.2])

# instantiate the base classifier.
lr = LogisticRegression(maxIter=10, tol=1E-6, fitIntercept=True)

# instantiate the One Vs Rest Classifier.
ovr = OneVsRest(classifier=lr)


# train the multiclass model.
ovrModel = ovr.fit(train)
lrm = lr.fit(train)

# score the model on test data.
predictions = ovrModel.transform(test)
predictions2 = lrm.transform(test)

predictions.show(6)
predictions2.show(6)

Upvotes: 0

Views: 1136

Answers (1)

Manoj Singh
Manoj Singh

Reputation: 1737

I don't think you can access the probabilities(confidence) vector because it takes the max of the confidence and drops the confidence vector. To test, you can make a copy of the class and modify it and remove the .drop(accColName)

http://spark.apache.org/docs/2.0.1/api/python/_modules/pyspark/ml/classification.html

# output the index of the classifier with highest confidence as prediction
labelUDF = udf(
    lambda predictions: float(max(enumerate(predictions), key=operator.itemgetter(1))[0]),
    DoubleType())

# output label and label metadata as prediction
return aggregatedDataset.withColumn(
    self.getPredictionCol(), labelUDF(aggregatedDataset[accColName])).drop(accColName)

Upvotes: 1

Related Questions