spark-core-map-01-unique-dates.scala

val orders = sc.textFile("/public/retail_db/orders")

orders.
  map(o => o.split(",")(1)).
  take(10).
  foreach(println)
  
orders.
  map(o => o.split(",")(1)).
  distinct.
  collect.
  foreach(println)

spark-core-map-02-unique-statuses.scala

orders.
  map(o => o.split(",")(3)).
  take(10).
  foreach(println)
  
orders.
  map(o => o.split(",")(3)).
  distinct.
  collect.
  foreach(println)

spark-core-map-03-order_id-and-dates.scala

orders.
  map(o => (o.split(",")(0).toInt, o.split(",")(1))).
  take(10).
  foreach(println)

spark-core-map-04-order_id-and-months.scala

orders.
  map(o => (o.split(",")(0).toInt, o.split(",")(1).substring(0, 7).replace("-", "").toInt)).
  take(10).
  foreach(println)