[PYTHON] Processing order when chaining when in PySpark

When I was writing code to chain when in PySpark

"Does this take precedence over the previously written when, just like SQL? " "Because it is a method chain, will it be overwritten by the when written at the end? "

I was worried, so I actually wrote the verification code and examined it.

Dummy data

df = spark.createDataFrame([(1,),(2,),(3,)], schema=('val',))
display(df)

val
1
2
3

For Spark SQL

#Registered as a temporary table to touch from Spark SQL
df.registerTempTable('tmp')

SELECT
  val,
  CASE
    WHEN val <= 1 THEN 'label_1'
    WHEN val <= 2 THEN 'label_2'
    ELSE 'label_3'
  END AS label
FROM tmp

val	label
1	label_1
2	label_2
3	label_3

In the case of SQL, of course, the condition of WHEN written earlier takes precedence.

For PySpark

from pyspark.sql import functions as F

df_label = df.withColumn('label',
    F.when(F.col('val') <= 1, 'label_1')
     .when(F.col('val') <= 2, 'label_2')
     .otherwise('label_3')
)
display(df_label)

val	label
1	label_1
2	label_2
3	label_3

Even when when is chained in PySpark, it seems that the condition of when written earlier has priority as in Spark SQL.