Iterable object from: local file system (Windows)¶
Our file is already residing on the windows file system, so we will start our url from the file:// intiator. Let’s instantiate the ObjectStorageDatast. The type of cal_housing object is osds.utils.ObjectStorageDataset.
from osds.utils import ObjectStorageDataset
from torch.utils.data import DataLoader
cal_housing = ObjectStorageDataset("file://C:\\Users\\laayt\\OneDrive\\Pictures\\california_housing_test.csv", batch_size = 10)
type(cal_housing)
>> osds.utils.ObjectStorageDataset
Batch object type is torch.Tensor. The batch size is 10, and number of columns are 9, so the shape of would be [10, 9].
batch = next(iter(DataLoader(cal_housing)))
type(batch)
>> torch.Tensor
batch.shape
>> torch.Size([1, 10, 9])
We also split database based on labels and features. Labels has only one row and 9 columns, so the shape of labels would be [1, 9]. The shape of features of each batch would be [9, 9].
labels, features = batch[:, 0], batch[:, 1:]
labels.shape
>> torch.Size([1, 9])
features.shape
>> torch.Size([1, 9, 9])