identify the most recent record different from the current value in a PySpark dataframe

Question

I have a PySpark dataframe, where each user has a certain status at a point in time, as in the dummy data below

    --------------------------
    |user_id| s