pyspark meandays calculation with respect to pandas code

Question

#Pandas code

temp = df_merge[[\'subscription_id\', \'cancelleddate\', \'subscriptionstartdate\', \'termenddate\']].drop_duplicates()
df_merge