Spark – How to Change Column Type?

To change the Spark SQL DataFrame column type from one data type to another data type you should use cast() function of Column class, you can use this on withColumn(), select(), selectExpr(), and SQL expression. Note that the type which you want to convert to should be a subclass of DataType class or a string representing the type.

Key points:

• cast() – cast() is a function from Column class that is used to convert the column into the other datatype.
• This function takes the argument string representing the type you wanted to convert or any type that is a subclass of DataType (Below mentioned the types you could cast).
• When Spark unable to convert into a specific type, it returns a null value.
• Spark SQL takes the different syntax DOUBLE(String column) to cast types.

ArrayType, BinaryType, BooleanType, CalendarIntervalType, DateType, HiveStringType, MapType, NullType, NumericType, ObjectType, StringType, StructType, TimestampType

import org.apache.spark.sql.functions.col
import org.apache.spark.sql.types.IntegerType

// Convert String to Integer Type
df.withColumn("salary",col("salary").cast(IntegerType))
df.withColumn("salary",col("salary").cast("int"))
df.withColumn("salary",col("salary").cast("integer"))

// Using select
df.select(col("salary").cast("int").as("salary"))

// Using selectExpr()
  df.selectExpr("cast(salary as int) salary","isGraduated")
  df.selectExpr("INT(salary)","isGraduated")

// Using with spark.sql()
spark.sql("SELECT INT(salary),BOOLEAN(isGraduated),gender from CastExample")
spark.sql("SELECT cast(salary as int) salary, BOOLEAN(isGraduated),gender from CastExample")

Let’s see some examples here using Scala snippet.

1. Change Column Type Example

First, let’s create DataFrame.

// Change Column Type Example
import org.apache.spark.sql.{Row, SparkSession}
import org.apache.spark.sql.types._
val spark: SparkSession = SparkSession.builder()
    .master("local[1]")
    .appName("SparkByExamples.com")
    .getOrCreate()

val simpleData = Seq(Row("James",34,"2006-01-01","true","M",3000.60),
    Row("Michael",33,"1980-01-10","true","F",3300.80),
    Row("Robert",37,"06-01-1992","false","M",5000.50)
  )

val simpleSchema = StructType(Array(
    StructField("firstName",StringType,true),
    StructField("age",IntegerType,true),
    StructField("jobStartDate",StringType,true),
    StructField("isGraduated", StringType, true),
    StructField("gender", StringType, true),
    StructField("salary", DoubleType, true)
  ))

val df = spark.createDataFrame(
     spark.sparkContext.parallelize(simpleData),simpleSchema)
df.printSchema()
df.show(false)

Outputs:

// Outputs:
root
 |-- firstName: string (nullable = true)
 |-- age: integer (nullable = true)
 |-- jobStartDate: string (nullable = true)
 |-- isGraduated: string (nullable = true)
 |-- gender: string (nullable = true)
 |-- salary: double (nullable = true)

+---------+---+------------+-----------+------+------+
|firstName|age|jobStartDate|isGraduated|gender|salary|
+---------+---+------------+-----------+------+------+
|James    |34 |2006-01-01  |true       |M     |3000.6|
|Michael  |33 |1980-01-10  |true       |F     |3300.8|
|Robert   |37 |06-01-1992  |false      |M     |5000.5|
+---------+---+------------+-----------+------+------+

2. Change Column Type using withColumn() and cast()

To convert the data type of a DataFrame column, Use withColumn() with the original column name as a first argument and for the second argument apply the casting method cast() with DataType on the column.

Below Spark, snippet changes DataFrame column, age to String (StringType), isGraduated column from String to Boolean (BooleanType) and ‘jobStartDate‘ column from String to DateType.

// Change Column Type using withColumn() and cast()
import org.apache.spark.sql.functions._
val df2 = df.withColumn("age",col("age").cast(StringType))
    .withColumn("isGraduated",col("isGraduated").cast(BooleanType))
    .withColumn("jobStartDate",col("jobStartDate").cast(DateType))
df2.printSchema()

Outputs:

// Outputs:
root
 |-- age: string (nullable = true)
 |-- isGraduated: boolean (nullable = true)
 |-- jobStartDate: date (nullable = true)

3. Using select() to Change Data Type

Below example cast’s selected columns using select() transformation. When you have many columns on DataFrame and wanted to cast selected columns this comes in handy.

// Using select() to Change Data Type
val cast_df = df.select(df.columns.map {
    case column@"age" =>
      col(column).cast("String").as(column)
    case column@"salary" =>
      col(column).cast("String").as(column)
    case column =>
      col(column)
  }: _*)

cast_df.printSchema()

col(column).cast("String").as(column)) will be converted as CAST(YEAR AS STRING) AS YEAR. Yields below output.

// Output:
root
 |-- firstName: string (nullable = true)
 |-- age: string (nullable = true)
 |-- jobStartDate: string (nullable = true)
 |-- isGraduated: string (nullable = true)
 |-- gender: string (nullable = true)
 |-- salary: string (nullable = true)

This example has been shared by @sriramrimmalapudi9gmail-com

4. Using selectExpr() to Change Data Type

Let’s use selectExpr() to convert spark DataFrame column age back to an integer, isGraduated from boolean to string and jobStartDate from date to String.

// Using selectExpr() to Change Data Type
val df3 = df2.selectExpr("cast(age as int) age",
    "cast(isGraduated as string) isGraduated",
    "cast(jobStartDate as string) jobStartDate")
df3.printSchema()
df3.show(false)

// Output:
root
 |-- age: integer (nullable = true)
 |-- isGraduated: string (nullable = true)
 |-- jobStartDate: string (nullable = true)

4. Using SQL Expression to Convert

We can also use SQL expression to change the spark DataFrame column type.

// Using SQL Expression to Convert
df3.createOrReplaceTempView("CastExample")
val df4 = spark.sql("SELECT STRING(age),BOOLEAN(isGraduated),
        DATE(jobStartDate) from CastExample")
df4.printSchema()
df4.show(false)

Outputs:

// Outputs:
root
 |-- age: string (nullable = true)
 |-- isGraduated: boolean (nullable = true)
 |-- jobStartDate: date (nullable = true)

5. Complete Example of Casting DataFrame Column

// Complete Example of Casting DataFrame Column 
package com.sparkbyexamples.spark.dataframe

import org.apache.spark.sql.{Row, SparkSession}
import org.apache.spark.sql.types._
import org.apache.spark.sql.functions._

object CastColumnType extends App{
  val spark: SparkSession = SparkSession.builder()
    .master("local[1]")
    .appName("SparkByExamples.com")
    .getOrCreate()

  val simpleData = Seq(Row("James",34,"2006-01-01","true","M",3000.60),
    Row("Michael",33,"1980-01-10","true","F",3300.80),
    Row("Robert",37,"06-01-1992","false","M",5000.50)
  )

  val simpleSchema = StructType(Array(
    StructField("firstName",StringType,true),
    StructField("age",IntegerType,true),
    StructField("jobStartDate",StringType,true),
    StructField("isGraduated", StringType, true),
    StructField("gender", StringType, true),
    StructField("salary", DoubleType, true)
  ))

  val df = spark.createDataFrame(
     spark.sparkContext.parallelize(simpleData),simpleSchema)
  df.printSchema()
  df.show(false)

  val df2 = df.withColumn("age",col("age").cast(StringType))
    .withColumn("isGraduated",col("isGraduated").cast(BooleanType))
    .withColumn("jobStartDate",col("jobStartDate").cast(DateType))
  df2.printSchema()

  val df3 = df2.selectExpr("cast(age as int) age",
    "cast(isGraduated as string) isGraduated",
    "cast(jobStartDate as string) jobStartDate")
  df3.printSchema()
  df3.show(false)

  df3.createOrReplaceTempView("CastExample")
  val df4 = spark.sql("SELECT STRING(age),BOOLEAN(isGraduated), " +
    "DATE(jobStartDate) from CastExample")
    df4.printSchema()
    df4.show(false)

  val cast_df = df.select(df.columns.map {
    case column@"age" =>
      col(column).cast("String").as(column)
    case column@"salary" =>
      col(column).cast("String").as(column)
    case column =>
      col(column)
  }: _*)

  cast_df.printSchema()
}

This example is also available at GitHub for reference.

Happy Learning !!

Related Articles

• Spark Check String Column Has Numeric Values
• Spark Using Length/Size Of a DataFrame Column
• Spark – Get Size/Length of Array & Map Column
• Spark Check Column Data Type is Integer or String
• Spark Merge Two DataFrames with Different Columns or Schema
• Spark Get DataType & Column Names of DataFrame
• Spark from_json() – Convert JSON Column to Struct, Map or Multiple Columns
• Spark DataFrame – Fetch More Than 20 Rows & Column Full Value