Practical Predictive Analytics

Comparing train and test decision trees

There is no prediction method for decision trees. However, we can still generate trees from both test and training datasets and compare them for reasonableness.

Using our diabetes example, we can first run a Pyspark decision tree on the training data. Again, our outcome variable is the first column (#0), which is referenced in the LabeledPoint reference, and the features are designated by all the variables that follow (1:):

%python  df = spark.sql("SELECT outcome, age, mass, triceps, pregnant, glucose, pressure, insulin, pedigree FROM global_temp.df_view") #sqlDF.show() rdd1 = df.rdd.map(lambda line:LabeledPoint(line[0],[line[1:]])) model_train = DecisionTree.trainClassifier(rdd1,numClasses=2,maxDepth=2,maxBins=32, ...