Is it correct that the maximum number of check-ins is set to 200 when bringing a highly similar trajectory from train data (including all users' data), and 600 when bringing a past trajectory of the target user? If you have some experiment results in case of using only target user's data, could you share the result here?
Thank you for pointing out the mistake. Yes, it should be shape[0].
The maximum number of check-ins is for all historical data. The result of using only the target user's data is reported as the second last row in Table 3.
Thank you for sharing your code. I have two questions.
To get the length of the most recent trajectory, is shape[0] correct?
num_traj = user_trajectory_data.shape[-1]
https://github.com/neolifer/LLM4POI/blob/7fc9e56f37fcf3b7b62fdba7227e1c6a6b3b9a77/preprocessing/to_nextpoi_kqt.py#L32C1
Is it correct that the maximum number of check-ins is set to 200 when bringing a highly similar trajectory from train data (including all users' data), and 600 when bringing a past trajectory of the target user? If you have some experiment results in case of using only target user's data, could you share the result here?