Classification - SparkML vs SynapseML

In this article, you perform the same classification task in two different ways: once using plain pyspark and once using the synapseml library. The two methods yield the same performance, but highlights the simplicity of using synapseml compared to pyspark.

The task is to predict whether a customer's review of a book sold on Amazon is good (rating > 3) or bad based on the text of the review. You accomplish it by training LogisticRegression learners with different hyperparameters and choosing the best model.

Setup

Import necessary Python libraries and get a spark session.

Read the data

Download and read in the data.

rawData = spark.read.parquet(
    "wasbs://publicwasb@mmlspark.blob.core.windows.net/BookReviewsFromAmazon10K.parquet"
)
rawData.show(5)

Extract features and process data

Real data is more complex than the above dataset. It's common for a dataset to have features of multiple types, such as text, numeric, and categorical. To illustrate how difficult it's to work with these datasets, add two numerical features to the dataset: the word count of the review and the mean word length.

from pyspark.sql.functions import udf
from pyspark.sql.types import *


def wordCount(s):
    return len(s.split())


def wordLength(s):
    import numpy as np

    ss = [len(w) for w in s.split()]
    return round(float(np.mean(ss)), 2)


wordLengthUDF = udf(wordLength, DoubleType())
wordCountUDF = udf(wordCount, IntegerType())

from synapse.ml.stages import UDFTransformer

wordLength = "wordLength"
wordCount = "wordCount"
wordLengthTransformer = UDFTransformer(
    inputCol="text", outputCol=wordLength, udf=wordLengthUDF
)
wordCountTransformer = UDFTransformer(
    inputCol="text", outputCol=wordCount, udf=wordCountUDF
)

from pyspark.ml import Pipeline

data = (
    Pipeline(stages=[wordLengthTransformer, wordCountTransformer])
    .fit(rawData)
    .transform(rawData)
    .withColumn("label", rawData["rating"] > 3)
    .drop("rating")
)

data.show(5)

Classify using pyspark

To choose the best LogisticRegression classifier using the pyspark library, we need to explicitly perform the following steps:

1. Process the features:
   • Tokenize the text column
   • Hash the tokenized column into a vector using hashing
   • Merge the numeric features with the vector
2. Process the label column: cast it into the proper type.
3. Train multiple LogisticRegression algorithms on the train dataset with different hyperparameters
4. Compute the area under the ROC curve for each of the trained models and select the model with the highest metric as computed on the test dataset
5. Evaluate the best model on the validation set

from pyspark.ml.feature import Tokenizer, HashingTF
from pyspark.ml.feature import VectorAssembler

# Featurize text column
tokenizer = Tokenizer(inputCol="text", outputCol="tokenizedText")
numFeatures = 10000
hashingScheme = HashingTF(
    inputCol="tokenizedText", outputCol="TextFeatures", numFeatures=numFeatures
)
tokenizedData = tokenizer.transform(data)
featurizedData = hashingScheme.transform(tokenizedData)

# Merge text and numeric features in one feature column
featureColumnsArray = ["TextFeatures", "wordCount", "wordLength"]
assembler = VectorAssembler(inputCols=featureColumnsArray, outputCol="features")
assembledData = assembler.transform(featurizedData)

# Select only columns of interest
# Convert rating column from boolean to int
processedData = assembledData.select("label", "features").withColumn(
    "label", assembledData.label.cast(IntegerType())
)

from pyspark.ml.evaluation import BinaryClassificationEvaluator
from pyspark.ml.classification import LogisticRegression

# Prepare data for learning
train, test, validation = processedData.randomSplit([0.60, 0.20, 0.20], seed=123)

# Train the models on the 'train' data
lrHyperParams = [0.05, 0.1, 0.2, 0.4]
logisticRegressions = [
    LogisticRegression(regParam=hyperParam) for hyperParam in lrHyperParams
]
evaluator = BinaryClassificationEvaluator(
    rawPredictionCol="rawPrediction", metricName="areaUnderROC"
)
metrics = []
models = []

# Select the best model
for learner in logisticRegressions:
    model = learner.fit(train)
    models.append(model)
    scoredData = model.transform(test)
    metrics.append(evaluator.evaluate(scoredData))
bestMetric = max(metrics)
bestModel = models[metrics.index(bestMetric)]

# Get AUC on the validation dataset
scoredVal = bestModel.transform(validation)
print(evaluator.evaluate(scoredVal))

Classify using SynapseML

The steps needed with synapseml are simpler:

1. The TrainClassifier Estimator featurizes the data internally, as long as the columns selected in the train, test, validation dataset represent the features

2. The FindBestModel Estimator finds the best model from a pool of trained models by finding the model that performs best on the test dataset given the specified metric

3. The ComputeModelStatistics Transformer computes the different metrics on a scored dataset (in our case, the validation dataset) at the same time

from synapse.ml.train import TrainClassifier, ComputeModelStatistics
from synapse.ml.automl import FindBestModel

# Prepare data for learning
train, test, validation = data.randomSplit([0.60, 0.20, 0.20], seed=123)

# Train the models on the 'train' data
lrHyperParams = [0.05, 0.1, 0.2, 0.4]
logisticRegressions = [
    LogisticRegression(regParam=hyperParam) for hyperParam in lrHyperParams
]
lrmodels = [
    TrainClassifier(model=lrm, labelCol="label", numFeatures=10000).fit(train)
    for lrm in logisticRegressions
]

# Select the best model
bestModel = FindBestModel(evaluationMetric="AUC", models=lrmodels).fit(test)


# Get AUC on the validation dataset
predictions = bestModel.transform(validation)
metrics = ComputeModelStatistics().transform(predictions)
print(
    "Best model's AUC on validation set = "
    + "{0:.2f}%".format(metrics.first()["AUC"] * 100)
)