I have a simple pyspark dataframe like this:
------------------------------------------------------- | timestamp_local|timestamp_utc|device|key |value| --