PySpark ML: Get best set of parameters from TrainValidationSplit

[Aakash Basu <aa...@gmail.com>]

Hi,

I am running a Random Forest model on a dataset using hyper parameter
tuning with Spark's paramGrid and Train Validation Split.

Can anyone tell me how to get the best set for all the four parameters?

I used:

model.bestModel()
model.metrics()


But none of them seem to work.


Below is the code chunk:

paramGrid = ParamGridBuilder() \
        .addGrid(rf.numTrees, [50, 100, 150, 200]) \
        .addGrid(rf.maxDepth, [5, 10, 15, 20]) \
        .addGrid(rf.minInfoGain, [0.001, 0.01, 0.1, 0.6]) \
        .addGrid(rf.minInstancesPerNode, [5, 15, 30, 50, 100]) \
        .build()

tvs = TrainValidationSplit(estimator=pipeline,
                           estimatorParamMaps=paramGrid,
                           evaluator=MulticlassClassificationEvaluator(),
                           # 80% of the data will be used for
training, 20% for validation.
                           trainRatio=0.8)

model = tvs.fit(trainingData)

predictions = model.transform(testData)

evaluator = MulticlassClassificationEvaluator(
        labelCol="label", predictionCol="prediction", metricName="accuracy")
accuracy = evaluator.evaluate(predictions)
print("Accuracy = %g" % accuracy)
print("Test Error = %g" % (1.0 - accuracy))


Any help?


Thanks,
Aakash.

Re: PySpark ML: Get best set of parameters from TrainValidationSplit

[Bryan Cutler <cu...@gmail.com>]

Hi Aakash,

First you will want to get the the random forest model stage from the best
pipeline model result, for example if RF is the first stage:

rfModel = model.bestModel.stages[0]

Then you can check the values of the params you tuned like this:

rfModel.getNumTrees

On Mon, Apr 16, 2018 at 7:52 AM, Aakash Basu <aa...@gmail.com>
wrote:

> Hi,
>
> I am running a Random Forest model on a dataset using hyper parameter
> tuning with Spark's paramGrid and Train Validation Split.
>
> Can anyone tell me how to get the best set for all the four parameters?
>
> I used:
>
> model.bestModel()
> model.metrics()
>
>
> But none of them seem to work.
>
>
> Below is the code chunk:
>
> paramGrid = ParamGridBuilder() \
>         .addGrid(rf.numTrees, [50, 100, 150, 200]) \
>         .addGrid(rf.maxDepth, [5, 10, 15, 20]) \
>         .addGrid(rf.minInfoGain, [0.001, 0.01, 0.1, 0.6]) \
>         .addGrid(rf.minInstancesPerNode, [5, 15, 30, 50, 100]) \
>         .build()
>
> tvs = TrainValidationSplit(estimator=pipeline,
>                            estimatorParamMaps=paramGrid,
>                            evaluator=MulticlassClassificationEvaluator(),
>                            # 80% of the data will be used for training, 20% for validation.
>                            trainRatio=0.8)
>
> model = tvs.fit(trainingData)
>
> predictions = model.transform(testData)
>
> evaluator = MulticlassClassificationEvaluator(
>         labelCol="label", predictionCol="prediction", metricName="accuracy")
> accuracy = evaluator.evaluate(predictions)
> print("Accuracy = %g" % accuracy)
> print("Test Error = %g" % (1.0 - accuracy))
>
>
> Any help?
>
>
> Thanks,
> Aakash.
>