def test_generate_subdataset(self, method="train_test", by=0.8):
"""
Creates a Project and a Dataset and links each other. Then creates a SubDataset out of the created Dataset.
The subdataset are created using method and by paramenters
Asserts
-------
- Indicies are not None
"""
Project(path=self.path_to_test_dir, name=self.aux_project_name)
ds = Dataset.read_file(path=self.path_to_dataset)
ds.save()
sbds = SubDataset(dataset=ds, method=method, by=by)
self.assertIsNotNone(sbds.indices)
return sbds
def test_create_dataset(self):
"""
Takes a dataset file and constructs a new dataset
Asserts
-------
- If the object created is an instance of a Dataset object
- If the datasource attribute is an instance of a FileDataset object
Returns
-------
A Dataset object instance created from the dataset file
"""
Project(name=self.aux_project_name, path=self.path_to_test_dir)
ds = Dataset.read_file(path=self.path_to_dataset,
first_line_heading=False)
self.assertIsInstance(ds, Dataset)
self.assertIsInstance(ds.datasource, FileDatasource)
return ds
from driftai.data.dataset import Dataset
from driftai.project import Project
from driftai.approach import Approach
from driftai import set_project_path
ds = Dataset.from_dir(
r"C:\cygwin64\home\fcgr\code\driftai\examples\mnst_digit_classification_old\data"
)
proj_path = r"C:\cygwin64\home\fcgr\code\driftai\test\my_tests"
proj_name = "a"
if Path("{}\{}".format(proj_path, proj_name)).exists():
shutil.rmtree(str(Path("{}\{}".format(proj_path, proj_name)).absolute()))
p = Project(name="a", path=r"C:\cygwin64\home\fcgr\code\driftai\test\my_tests")
set_project_path(p.path)
info = ds.get_info()
ds.save()
sbs = ds.generate_subdataset(method="k_fold", by=5)
sbs.save()
from PIL import Image
import numpy as np
#tr_data = sbs.get_train_data("A", loader=lambda x: np.asarray(Image.open(x)).reshape(-1))
#ts_data = sbs.get_test_data("A", loader=lambda x: np.asarray(Image.open(x)))
a = Approach(project=p, name="my_approach", subdataset=sbs)
def test_automatically_detect_regression(self):
Project(name=self.aux_project_name, path=self.path_to_test_dir)
ds = Dataset.read_file("test/resources/housing.csv",
label="median_house_value")
self.assertEqual(ds.problem_type, "regression")
def test_automatically_detect_binary_clf(self):
Project(name=self.aux_project_name, path=self.path_to_test_dir)
ds = Dataset.read_file("test/resources/titanic.csv", label="Survived")
self.assertEqual(ds.problem_type, "binary_clf")
def test_automatically_detect_clf(self):
Project(name=self.aux_project_name, path=self.path_to_test_dir)
ds = Dataset.read_file("test/resources/Iris.csv")
self.assertEqual(ds.problem_type, "clf")