Transformations vs actions in Apache Spark - Practice Questions

Challenge - 5 Problems

❓ Predict Output

intermediate

Understanding lazy evaluation in Spark

What will be the output of this Spark code snippet?

Apache Spark

val rdd = sc.parallelize(Seq(1, 2, 3, 4))
val mapped = rdd.map(x => x * 2)
// No action called yet
mapped.count()

A 4

B 8

C No output, code runs lazily without action

D Throws an error because no action is called

💡 Hint

Remember that transformations are lazy and actions trigger computation.

The map is a transformation and does not execute immediately. The count() is an action that triggers the computation and returns the number of elements, which is 4.

❓ data_output

intermediate

Result of chained transformations and action

Given this Spark code, what is the output of the collect() action?

Apache Spark

val rdd = sc.parallelize(Seq(1, 2, 3, 4, 5))
val filtered = rdd.filter(_ % 2 == 0)
val mapped = filtered.map(_ * 10)
mapped.collect()

A [10, 20]

B [2, 4]

C [1, 3, 5]

D [20, 40]

💡 Hint

Filter keeps even numbers, then map multiplies by 10.

The filter keeps 2 and 4, then map multiplies them by 10 resulting in 20 and 40.

🔧 Debug

advanced

Why does this Spark job not run?

Consider this Spark code snippet. Why does it not produce any output or run any job?

Apache Spark

val rdd = sc.parallelize(Seq(1, 2, 3))
rdd.map(_ * 2)

A Because <code>map</code> is a transformation and no action is called to trigger execution

B Because <code>map</code> is an action and should run immediately

C Because the RDD is empty

D Because SparkContext is not initialized

💡 Hint

Think about what triggers Spark jobs to run.

Transformations like map are lazy and do not execute until an action is called.

🧠 Conceptual

advanced

Difference between transformations and actions

Which statement correctly describes the difference between transformations and actions in Spark?

A Transformations immediately compute results; actions only define computations

B Actions are lazy and transformations trigger execution

C Transformations define a new RDD without executing; actions trigger execution and return results

D Both transformations and actions trigger execution immediately

💡 Hint

Recall which operations cause Spark to run jobs.

Transformations build a computation plan and are lazy. Actions trigger execution and return results.

🚀 Application

expert

Optimizing Spark job with transformations and actions

You have this Spark code. Which option will minimize the number of Spark jobs executed?

Apache Spark

val rdd = sc.parallelize(Seq(1, 2, 3, 4, 5))
val mapped = rdd.map(_ * 2)
val filtered = mapped.filter(_ > 5)
val countResult = filtered.count()
val collectResult = filtered.collect()

A Call count and collect without caching; Spark will optimize automatically

B Cache the filtered RDD before calling count and collect

C Call collect first, then count; order does not matter

D Use two separate RDDs for count and collect to avoid recomputation

💡 Hint

Think about how caching affects repeated actions on the same RDD.

Caching the filtered RDD stores its data in memory, so the second action does not recompute the transformations again, reducing Spark jobs.