DataFrames_detailed.txt

df.select('order_id', 'order_date').show() 	#some functions take column names as literals.
df.select(df.order_id, df.order_date).show()#some functions take dfname with column name.
So, be careful while using, SELECT accepts in both the ways.

df.select('*') # to get all the data.
from pyspark.sql import functions as f
orders.select('id', 'date', f.date_format(orders.date, "YYYYMM").alias('date_month')).show()

selectExpr # lets you sql type sytanx

orders.selectExpr('date_format(order_date, "YYYYMM") as new').show() # you don't need import pyspark.sql.functions in this case.

BY DEFAULT join uses 200 thread to make it faster but based on your data you can use following
	spark.conf.set('spark.sql.shuffle.partions', '1')