Made the project 2 clear. We are looking for a seq model that can replace the gym.step
Removing outliers only if they are not in the middle of an episode (trip). Otherwise consider impute
Countdown. Based on the wall clock value and schedule of the ferry. Create how long it passed from the scheduled start time (not the trip seq). Suggestion, create expected arrival time, then subtract it (in parallel) from the datetime column.
Solution 1: A workaround is to store the string lists as numpy byte objects because they only have 1 refcount. (check for other solutions yourself!) This will create a robust reproducable trips from datastream. You can check if you still have memory leak using memory profiler (https://upaspro.com/how-to-check-for-memory-leak-in-python/)
Solution2: Do not manipulate trips based on hard coded IDs. Try to select trips dynamically.
Action items:
Create one-week plan for your project with the estimated finish time
High level plan
@tzhao-ooc Expected to finish All data prepration and feature selection/engineering by next meeting (June 7)
@jnqian99 Finish first Seq model for a gym enviroenmnt (half cheetah) by next meeting (June 7)
Solution 1: A workaround is to store the string lists as numpy byte objects because they only have 1 refcount. (check for other solutions yourself!) This will create a robust reproducable trips from datastream. You can check if you still have memory leak using memory profiler (https://upaspro.com/how-to-check-for-memory-leak-in-python/) Solution2: Do not manipulate trips based on hard coded IDs. Try to select trips dynamically.
Action items:
High level plan
@tzhao-ooc Expected to finish All data prepration and feature selection/engineering by next meeting (June 7)
@jnqian99 Finish first Seq model for a gym enviroenmnt (half cheetah) by next meeting (June 7)