Просто используйте to_date
в формате w/yyyy
:
df = spark.createDataFrame([(1, "18/2020")], ['id', 'week_year'])
df.withColumn("date", to_date(col("week_year"), "w/yyyy")).show()
#+---+---------+----------+
#| id|week_year| date|
#+---+---------+----------+
#| 1| 18/2020|2020-04-26|
#+---+---------+----------+