使用PySpark Dataframe平均超过2000个值

I have a PySpark dataframe with about a billion rows. I want to average over every 2000 values, like average of rows with indeces 0-1999, average of rows with indeces 2000-3999, and so on. How do I do this? Alternatively, I could also average 10 values for every 2000, like average of rows with indeces 0-9, average of rows with indeces 2000-2009, and so on. The purpose of this is to downsample the data. I don't currently have an index row, so if I need this, how do I do it?I have a PySpark dataframe with about a billion