Sprint #7 (June 15)

[pagand]

Feedback:

• Only model transition data and valid trips
• Dealing with outlier values: either impute or remove the whole trip if cannot impute
• Double check on mode
• Classify normal and adversarial situations (i.e, rerouting, significant drop in velocity, etc.)
• Check episodes in time series / reinforcement learning to get batch

Action items:

• Finalize the preprocessing stage
• Implement one simple model