Dataset about cast HOT 5 CLOSED

I have the same question and look forward to the author's reply.

from cast.

need help too.

from cast.

Thank you for your interest in our work. To generate the 'peacor_adj.npy' file, you can use the code provided below.

def get_peacor_adj(data_path, threshold, save=False):
    # Load the dataset
    data = np.load(data_path + 'train.npz')['data']
    print("Data shape:", data.shape)
    
    # Compute the Pearson correlation coefficient matrix
    peacor = torch.corrcoef(torch.Tensor(data[...,0]).permute(1, 0))
    
    # Apply threshold
    peacor[peacor < threshold] = 0
    peacor[torch.eye(peacor.shape[0], dtype=bool)] = 0

    # Normalize the coefficients
    nonzero_peacor = peacor[peacor != 0]
    p_min, p_max = nonzero_peacor.min(), nonzero_peacor.max()
    peacor[peacor != 0] = (nonzero_peacor - p_min) / (p_max - p_min)

    # Visualization
    plt.figure(dpi=100)
    sns.heatmap(peacor)
    plt.show()
    
    # Save the result
    if save:
        np.save(data_path + 'peacor_adj.npy', peacor)

For customization, you can also consider implementing your own method for creating the edge attribute, which can be an alternative to the Pearson correlation or the Time-delayed Dynamic Time Warping (DTW) method used in our paper.

from cast.

Thank you for your interest in our work. To generate the 'peacor_adj.npy' file, you can use the code provided below.

def get_peacor_adj(data_path, threshold, save=False):
    # Load the dataset
    data = np.load(data_path + 'train.npz')['data']
    print("Data shape:", data.shape)
    
    # Compute the Pearson correlation coefficient matrix
    peacor = torch.corrcoef(torch.Tensor(data[...,0]).permute(1, 0))
    
    # Apply threshold
    peacor[peacor < threshold] = 0
    peacor[torch.eye(peacor.shape[0], dtype=bool)] = 0

    # Normalize the coefficients
    nonzero_peacor = peacor[peacor != 0]
    p_min, p_max = nonzero_peacor.min(), nonzero_peacor.max()
    peacor[peacor != 0] = (nonzero_peacor - p_min) / (p_max - p_min)

    # Visualization
    plt.figure(dpi=100)
    sns.heatmap(peacor)
    plt.show()
    
    # Save the result
    if save:
        np.save(data_path + 'peacor_adj.npy', peacor)

For customization, you can also consider implementing your own method for creating the edge attribute, which can be an alternative to the Pearson correlation or the Time-delayed Dynamic Time Warping (DTW) method used in our paper.

Thank you very much for your work, but I have the same problem with this issue, that is, there are some other files, how do I get these files, or is there a python script that can generate these files?
Issue Address: #2
Need files: such as "adj_mx_pems08.pkl", "time_dalay_attr.pkl", "sparse_adj.npy", "peacor_adj.npy" and "dist_adj.npy".

from cast.

如何获取数据集（/pems8/dataset.npy，peacor_adj.npy ....）？

Excuse me for the interruption.
Have you successfully run this model? I want to run it, but I'm missing some necessary data files . Could you please share how you handled this issue?
I would greatly appreciate it.

from cast.