I‘m a beginer for Python and need some help about comparing-trajectory-clustering-methods

[seljukgulcan]

Hello, Baoxin1997,

You can look at https://github.com/Shathra/comparing-trajectory-clustering-methods/blob/master/demo/demo.ipynb for a start. If you can convert your data into the format described on the link (each trajectory should be (N, 2) shaped NumPy array), then the code should work. To do that, you may need to group your data by track ID and sort by timestamp. Please check out a NumPy quickstart tutorial if you don't know how to import, edit, sort, etc. an array using Numpy.

Also, euclidian distance doesn't make sense for polar coordinates. You may need to either use haversine distance or project your coordinates into a cartesian plane. The latter may not work if you are working on world-scale data. For more information, you can look at this wiki page: https://en.wikipedia.org/wiki/Geographic_coordinate_system#Grid_coordinates

[Baoxin1997]

Really thanks for your help! As you told me, I change the data in to NumPy array with MATLAB. So, how to change the code into read my local file? look forward your reply

[seljukgulcan]

Sorry, I'm not familiar with MATLAB. If you have a file that NumPy can import, you can import with numpy.load or numpy.loadtxt depending on the format of the file. Please refer to the corresponding documentation pages for more info.

[Baoxin1997]

Ok, I will try for it. Also, I make a dataset like you give me in demo, especially the 'cross.mat' one. If I want to use this file, how could I do for it? Because in demo code, you dwonload the dataset with a link not use a local file.

[seljukgulcan]

These are the lines you're looking for. Depending on your data, you may need to tweak some lines. After importing and preprocessing your data, traj_lst object should be a list of trajectories where each trajectory is represented by a (N, 2) NumPy array.

# Import dataset
traj_data = scipy.io.loadmat(filename)['tracks']

traj_lst = []
for data_instance in traj_data:
    traj_lst.append(np.vstack(data_instance[0]).T)

If it doesn't work please provide a sample output you get after you load your file with scipy.io.loadmat function.

Edit: 'track' key in the code might be something else for your data.

[Baoxin1997]

Really*N thanks!

I finish my work perfectly with your help. You are really kindly for me.

Let me thank you again!