In this Chapter, instead of using Spark shell, we will build two standalone Spark applications using Scala API: one for model preparation and the second for model deployment. In the case of Spark, the Spark application is a normal Scala application with a main method that serves as an entry point for execution. For example, here is a skeleton of application for model training:

object Chapter8 extends App {

val spark = SparkSession.builder()
     .master("local[*]")
     .appName("Chapter8")
     .getOrCreate()

val sc = spark.sparkContext
sc.setLogLevel("WARN")
script(spark, sc, spark.sqlContext)


def script(spark: SparkSession, sc: SparkContext, sqlContext: SQLContext): Unit = {
      // ...code of application
}
}
  

Moreover, we will try to extract parts, which can be shared between both applications, into a library. This will allow us to follow the DRY (do-not-repeat-yourself) principle:

object Chapter8Library {
    // ...code of library
  }