def _load_data(self): # pragma: no cover
(x_train, y_train), (x_test, y_test) = fashion_mnist.load_data()
train = Dataset(x_train, y_train, self.x_type, self.y_type)
test = Dataset(x_test, y_test, self.x_type, self.y_type)
return train, test, None
def _load_data(self): # pragma: no cover
(x_train, y_train), (x_test, y_test) = boston_housing.load_data()
train = Dataset(x_train, y_train, self.x_type, self.y_type)
test = Dataset(x_test, y_test, self.x_type, self.y_type)
return train, test, None
def _load_data(self):
train = Dataset(np.array([1]), np.array([10]), self.x_type, self.y_type)
test = Dataset(np.array([2]), np.array([20]), self.x_type, self.y_type)
validation = Dataset(
np.array([3]), np.array([30]), self.x_type, self.y_type
)
return train, test, validation
def __init__(self, parent: "QObject" = None):
super().__init__(parent)
self.max_row_batch_to_load = 100
self._dataset: "Dataset" = Dataset()
self._types: List[Optional[DataType]] = [
self._dataset.x_type,
self._dataset.y_type,
]
self._loaded_types: List[Dict[str, DataType]] = [
{
type(self._dataset.x_type).__name__: self._dataset.x_type
},
{
type(self._dataset.y_type).__name__: self._dataset.y_type
},
]
self._cached_data: List[List[Any]] = [[], []]
self._role_map = {
Qt.DisplayRole: self._display_role,
self.TypeRole: self._type_role,
}
def simple_categorical_dataset():
return Dataset(
x_data=np.array([0, 1, 2]),
y_data=np.array([0, 1, 2]),
x_type=Numeric(),
y_type=Categorical(["foo", "bar", "hue"]),
)
def simple_numeric_dataset():
return Dataset(
x_data=np.array([1, 2, 3, 4]),
y_data=np.array([10, 20, 30, 40]),
x_type=Numeric(),
y_type=Numeric(),
)
def test_dataset():
return Dataset(
x_data=np.array(
[np.array([1, 1, 1]),
np.array([2, 2, 2]),
np.array([3, 3, 3])]),
y_data=np.array([1, 2, 3]),
x_type=NumericArray(),
y_type=Numeric(),
)
def load(self, parent_dir: str) -> "Dataset":
"""Loads the dataset from the specified `dataset_description` object. A
`parent_dir` must be passed to resolve relative paths on the
`dataset_description`.
Returns:
The loaded dataset.
"""
x_type = self.get_x_type()
y_type = self.get_y_type()
# Dataset data (x, y) must be filled by subclasses overriding this method
return Dataset(x_type=x_type, y_type=y_type)
def _load_data(self): # pragma: no cover
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
train = Dataset(x_train, y_train, self.x_type, self.y_type)
test = Dataset(x_test, y_test, self.x_type, self.y_type)
train.y = np.array(
[train.y_type.convert_to_expected_format(i) for i in train.y])
test.y = np.array(
[test.y_type.convert_to_expected_format(i) for i in test.y])
return train, test, None
def empty_dataset():
return Dataset()