sinadabiri / deep-semi-supervised-gps-transport-mode Goto Github PK

Hi,

Is there any code regarding segmentation using ruptures?

Thank you.

Guys,

Does anyone managed to modify the preprocessing scripts here, so that we obtain also the trip ID?

I would like to do some post-processing task, that require the trip IDs for all segments in the test set.

Dear Dr. Dabiri

Firstly, thank you for your great job in Semi-Supervised Deep Learning Approach for Transportation Mode Identification Using GPS Trajectory Data and your sharing codes on Github.

Unfortunately, I got into a trouble when run '2-Label-Trajectory-All.py'.
I have tried to fix it myself but I failed.

Here are my environment and reproduce steps.
My environment

1. Ubuntu 18.04.2 LTS
2. Python 3.6.8 :: Anaconda, Inc.

Reproduce steps

1. I cloned your code. Then I downloaded GeoLife datasets and unzip it into directory "../Geolife Trajectories 1.3-Raw-All".
2. execute python 2-Label-Trajectory-All.py and an error occurs:

Traceback (most recent call last):
  2     File "2_Label_Trajectory_All.py", line 105, in <module>
  3     ¦   raise e
  4     File "2_Label_Trajectory_All.py", line 101, in <module>
  5     ¦   trajectory_all_user_with_labetory-All.pyk)
  6 IndexError: list index out of range

I would really appreciate it if you could give me an tip about solving this problem.
Thank you very very much.

The script 2_Label_Trajectory_All.py assumes that all the .plt files are sorted according to the timestamp(which is also the name) but it is not always the case. This causes problems when assigning the labels to correct trajectories.

I added the following line to sort the array that contains names of the .plt files.
`

    user_trajectories = os.listdir(trajectory_dir) 
    //sort all the .plt files in trajectory_dir
    user_trajectories = list(map(lambda sorted_plt : str(sorted_plt) + '.plt', sorted(list(map(lambda x: int(x.rstrip('.plt')), 
    user_trajectories)))))
    trajectory_one_user = []

`