def test_categorical_target(tmpdir):
train_df = TEST_DF_1.copy()
valid_df = TEST_DF_2.copy()
test_df = TEST_DF_2.copy()
for df in [train_df, valid_df, test_df]:
# change int label to string
df["label"] = df["label"].astype(str)
dm = TabularData(
train_df,
categorical_input=["category"],
numerical_input=["scalar_b", "scalar_b"],
target="label",
valid_df=valid_df,
test_df=test_df,
num_workers=0,
batch_size=1,
)
for dl in [
dm.train_dataloader(),
dm.val_dataloader(),
dm.test_dataloader()
]:
(cat, num), target = next(iter(dl))
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
def test_empty_inputs():
train_df = TEST_DF_1.copy()
with pytest.raises(RuntimeError):
TabularData.from_df(train_df,
numerical_cols=None,
categorical_cols=None,
target_col="label",
num_workers=0,
batch_size=1)
def test_empty_inputs():
train_data_frame = TEST_DF_1.copy()
with pytest.raises(RuntimeError):
TabularData.from_data_frame(
numerical_fields=None,
categorical_fields=None,
target_fields="label",
train_data_frame=train_data_frame,
num_workers=0,
batch_size=1,
)
def test_categorical_target(tmpdir):
train_data_frame = TEST_DF_1.copy()
val_data_frame = TEST_DF_2.copy()
test_data_frame = TEST_DF_2.copy()
for df in [train_data_frame, val_data_frame, test_data_frame]:
# change int label to string
df["label"] = df["label"].astype(str)
dm = TabularData.from_data_frame(
categorical_fields=["category"],
numerical_fields=["scalar_b", "scalar_b"],
target_fields="label",
train_data_frame=train_data_frame,
val_data_frame=val_data_frame,
test_data_frame=test_data_frame,
num_workers=0,
batch_size=1,
)
for dl in [dm.train_dataloader(), dm.val_dataloader(), dm.test_dataloader()]:
data = next(iter(dl))
(cat, num) = data[DefaultDataKeys.INPUT]
target = data[DefaultDataKeys.TARGET]
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
def test_from_csv(tmpdir):
train_csv = Path(tmpdir) / "train.csv"
val_csv = test_csv = Path(tmpdir) / "valid.csv"
TEST_DF_1.to_csv(train_csv)
TEST_DF_2.to_csv(val_csv)
TEST_DF_2.to_csv(test_csv)
dm = TabularData.from_csv(categorical_fields=["category"],
numerical_fields=["scalar_a", "scalar_b"],
target_fields="label",
train_file=str(train_csv),
val_file=str(val_csv),
test_file=str(test_csv),
num_workers=0,
batch_size=1)
for dl in [
dm.train_dataloader(),
dm.val_dataloader(),
dm.test_dataloader()
]:
data = next(iter(dl))
(cat, num) = data[DefaultDataKeys.INPUT]
target = data[DefaultDataKeys.TARGET]
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
def test_classification(tmpdir):
train_df = TEST_DF_1.copy()
val_df = TEST_DF_1.copy()
test_df = TEST_DF_1.copy()
data = TabularData.from_df(
train_df,
categorical_cols=["category"],
numerical_cols=["scalar_a", "scalar_b"],
target_col="label",
val_df=val_df,
test_df=test_df,
num_workers=0,
batch_size=2,
)
model = TabularClassifier(num_features=3, num_classes=2, embedding_sizes=data.emb_sizes)
trainer = pl.Trainer(fast_dev_run=True, default_root_dir=tmpdir)
trainer.fit(model, data)
def test_from_df(tmpdir):
train_df = TEST_DF_1.copy()
val_df = TEST_DF_2.copy()
test_df = TEST_DF_2.copy()
dm = TabularData.from_df(train_df,
categorical_cols=["category"],
numerical_cols=["scalar_b", "scalar_b"],
target_col="label",
val_df=val_df,
test_df=test_df,
num_workers=0,
batch_size=1)
for dl in [
dm.train_dataloader(),
dm.val_dataloader(),
dm.test_dataloader()
]:
(cat, num), target = next(iter(dl))
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
def test_tabular_data(tmpdir):
train_data_frame = TEST_DF_1.copy()
val_data_frame = TEST_DF_2.copy()
test_data_frame = TEST_DF_2.copy()
dm = TabularData.from_data_frame(
categorical_cols=["category"],
numerical_cols=["scalar_b", "scalar_b"],
target_col="label",
train_data_frame=train_data_frame,
val_data_frame=val_data_frame,
test_data_frame=test_data_frame,
num_workers=0,
batch_size=1,
)
for dl in [dm.train_dataloader(), dm.val_dataloader(), dm.test_dataloader()]:
data = next(iter(dl))
(cat, num) = data[DefaultDataKeys.INPUT]
target = data[DefaultDataKeys.TARGET]
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
def test_from_csv(tmpdir):
train_csv = Path(tmpdir) / "train.csv"
val_csv = test_csv = Path(tmpdir) / "valid.csv"
TEST_DF_1.to_csv(train_csv)
TEST_DF_2.to_csv(val_csv)
TEST_DF_2.to_csv(test_csv)
dm = TabularData.from_csv(train_csv=train_csv,
categorical_cols=["category"],
numerical_cols=["scalar_b", "scalar_b"],
target_col="label",
val_csv=val_csv,
test_csv=test_csv,
num_workers=0,
batch_size=1)
for dl in [
dm.train_dataloader(),
dm.val_dataloader(),
dm.test_dataloader()
]:
(cat, num), target = next(iter(dl))
assert cat.shape == (1, 1)
assert num.shape == (1, 2)
assert target.shape == (1, )
# See the License for the specific language governing permissions and
# limitations under the License.
from torchmetrics.classification import Accuracy, Precision, Recall
import flash
from flash.data.utils import download_data
from flash.tabular import TabularClassifier, TabularData
# 1. Download the data
download_data("https://pl-flash-data.s3.amazonaws.com/titanic.zip", "data/")
# 2. Load the data
datamodule = TabularData.from_csv(
["Sex", "Age", "SibSp", "Parch", "Ticket", "Cabin", "Embarked"],
["Fare"],
target_field="Survived",
train_file="./data/titanic/titanic.csv",
test_file="./data/titanic/test.csv",
val_split=0.25,
)
# 3. Build the model
model = TabularClassifier.from_data(
datamodule, metrics=[Accuracy(), Precision(),
Recall()])
# 4. Create the trainer
trainer = flash.Trainer(fast_dev_run=True)
# 5. Train the model
trainer.fit(model, datamodule=datamodule)
# limitations under the License.
from torchmetrics.classification import Accuracy, Precision, Recall
import flash
from flash.data.utils import download_data
from flash.tabular import TabularClassifier, TabularData
# 1. Download the data
download_data("https://pl-flash-data.s3.amazonaws.com/titanic.zip", "data/")
# 2. Load the data
datamodule = TabularData.from_csv(
target_col="Survived",
train_csv="./data/titanic/titanic.csv",
test_csv="./data/titanic/test.csv",
categorical_cols=[
"Sex", "Age", "SibSp", "Parch", "Ticket", "Cabin", "Embarked"
],
numerical_cols=["Fare"],
val_size=0.25,
)
# 3. Build the model
model = TabularClassifier.from_data(
datamodule, metrics=[Accuracy(), Precision(),
Recall()])
# 4. Create the trainer
trainer = flash.Trainer(fast_dev_run=True)
# 5. Train the model
trainer.fit(model, datamodule=datamodule)