Machine Learning with Spark - Second Edition

Step size

We will perform a similar analysis for step size in the following code:

Scala

val steps_param = Array(0.01, 0.025, 0.05, 0.1, 1.0) val intercept =false val i = 0 val results = new Array[String](5) val resultsMap = new scala.collection.mutable.HashMap[String, String] val dataset = new DefaultCategoryDataset() for(i <- 0 until steps_param.length) {   val step = steps_param(i)   val rmsle = LinearRegressionUtil.evaluate(train_data,          test_data,iterations,step,intercept)   resultsMap.put(step.toString,rmsle.toString)   dataset.addValue(rmsle, "RMSLE", step) } 

Output for the previous code is as follows:

    [1.7904244862988534, 1.4241062778987466, 1.3840130355866163,    1.4560061007109475, nan]

The plot for the preceding output is shown as ...