Dataset implementation for classification task?

[hillarylia]

Hi there,

I'm interested in time series classification. I noticed that there is no implementation for ClassificationDFDataset. Is there a workaround you might suggest? Otherwise, are there plans to implement classification?

[lkluo]

I have no idea why they dont implement classification. Perhaps they pretend no body cares time series classification. A quick solution is to modify on RegressionDFDataset:

from torch.utils.data import  Dataset

dataset = RegressionDFDataset(...)
dataset = ClassificationDataset(dataset)

class ClassificationDataset(Dataset):
    def __init__(self, data):
        self.data = data

    def __len__(self):
        return len(self.data)

    def __getitem__(self, index):
        ret = {
            "past_values": self.data[index]["past_values"],
            "target_values": self.data[index]["target_values"][0]
        }
        if self.data.datetime_col:
            ret["timestamp"] = self.data[index]["timestamp"]
        return ret

[hillarylia]

Thanks!