CODEWITHSUNDEEP
scalaapache-sparkapache-spark-sql
Sam
Sam

Reputation: 47

Spark Scala - Comparing columns values and then comparing result with another column

I have a data as below. Which contains, Product ID, Making cost, tax 1, tax 2, and MRP. So i have to create new Column deciding selling price of the item. Selling price show follow below condition

Example MC+TX1=70, MC+TX2=80, MRP = 100, then Selling price = 80

if MC+TX1 =170, MC+TX2 =80, MRP =100, then selling price = 80, since one of the MC+tx1 or MX+tx2 is smaller than MRP, then we will take that

MC+TX1=170, MC+TX2=180, MRP = 100, then Selling price = 100

 +------------+-------------+-------+-------+-----+--+
 | Product ID | Making Cost | Tax 1 | Tax 2 | MRP |  |
 +------------+-------------+-------+-------+-----+--+
 | 12345      | 50          | 20    | 30    | 100 |  |
 | 23456      | 50          | 60    | 30    | 100 |  |
 | 34567      | 50          | 60    | 70    | 100 |  |
 +------------+-------------+-------+-------+-----+--+

Desired Output

| Product ID | Making Cost | Tax 1 | Tax 2 | MRP | Selling Price |
|------------|-------------|-------|-------|-----|---------------|
| 12345      | 50          | 20    | 30    | 100 | 80            |
| 23456      | 50          | 60    | 30    | 100 | 80            |
| 34567      | 50          | 60    | 70    | 100 | 100           |

Upvotes: 1

Views: 54

Answers (2)

koiralo
koiralo

Reputation: 23119

You can use when for all the conditions as below

val df = spark.sparkContext.parallelize(Seq(
    (12345, 50, 20, 30, 100),
    (23456, 50, 60, 30, 100),
    (34567, 50, 60, 70, 100)
  ))
    .toDF("ID", "MC", "T1", "T2", "MRP")


  df.withColumn("SP",
    when((($"MC" + $"T1").gt($"MRP")) && (($"MC" + $"T2").gt($"MRP")), $"MRP")
      .otherwise(
        when((($"MC" + $"T1").lt($"MRP")) && (($"MC" + $"T2").lt($"MRP")), greatest(($"MC" + $"T1"), ($"MC" + $"T2")))
          .otherwise(
            when((($"MC" + $"T1").gt($"MRP")), $"MC" + $"T2")
              .otherwise($"MC" + $"T1")
          )
        )
  ).show(false)

OUTPUT:

+-----+---+---+---+---+---+
|ID   |MC |T1 |T2 |MRP|SP |
+-----+---+---+---+---+---+
|12345|50 |20 |30 |100|80 |
|23456|50 |60 |30 |100|80 |
|34567|50 |60 |70 |100|100|
+-----+---+---+---+---+---+

To make more efficient you can calculate all MC+T1 and M+T2 before so that you don't have to calculate every time and drop later if you don't want it.

Upvotes: 1

uh_big_mike_boi
uh_big_mike_boi

Reputation: 3470

You can do it with spark sql or with the dataframe API

df.withColumn("selling_price", 
  when((col("MC") + col("TX1") < col("MC") + col("TX2")) && (col("MC") + col("TX2") < col("MRP") , col("MC") + col("TX2"))     
 .when((col("MC") + col("TX1") > col("MRP")) && (col("MC") + col("TX2") > col("MRP") , col("MRP"))
 .otherwise(col("MC") + col("TX1")))

Upvotes: 0

Related Questions