def __init__(self, columns, bins=10):
super(ContinuousHistogramProfiler, self).__init__(columns)
self._bins = bins
self._np = maybe_dependency("numpy")
if self._np is None:
raise ImportError(
"numpy is not installed; try `pip install numpy`")
def as_dataframe(self):
"""
Returns this collection of Experiment Runs as a table.
Returns
-------
:class:`pandas.DataFrame`
"""
pd = importer.maybe_dependency("pandas")
if pd is None:
e = ImportError("pandas is not installed; try `pip install pandas`")
six.raise_from(e, None)
ids = []
data = []
for run in self:
run_data = {}
run_data.update({'hpp.'+k: v for k, v in run.get_hyperparameters().items()})
run_data.update({'metric.'+k: v for k, v in run.get_metrics().items()})
ids.append(run.id)
data.append(run_data)
columns = set()
for run_data in data:
columns.update(run_data.keys())
return pd.DataFrame(data, index=ids, columns=sorted(list(columns)))
def log_reference_data(self, X, Y, overwrite=False):
"""Log tabular reference data.
Parameters
----------
X : pd.DataFrame
Reference data inputs.
Y : pd.DataFrame
Reference data outputs.
overwrite : bool, default False
Whether to allow overwriting existing reference data.
"""
pd = importer.maybe_dependency("pandas")
if pd is None:
raise ImportError(
"pandas is not installed; try `pip install pandas`")
if isinstance(X, pd.Series):
X = X.to_frame()
if isinstance(Y, pd.Series):
Y = Y.to_frame()
if not isinstance(X, pd.DataFrame):
raise TypeError("`X` must be a DataFrame, not {}".format(type(X)))
if not isinstance(Y, pd.DataFrame):
raise TypeError("`Y` must be a DataFrame, not {}".format(type(Y)))
df = pd.DataFrame()
for c in X.columns:
df["input." + str(c)] = X[c]
for c in Y.columns:
df["output." + str(c)] = Y[c]
df["source"] = "reference"
df["model_version_id"] = self.id
tempf = tempfile.NamedTemporaryFile(suffix=".csv", delete=False)
try:
df.to_csv(tempf.name, encoding="utf-8", index=False)
self.log_artifact("reference_data",
tempf.name,
overwrite=overwrite)
finally:
os.remove(tempf.name)
def get_artifact(self, key):
"""
Gets the artifact with name `key` from this Model Version.
If the artifact was originally logged as just a filesystem path, that path will be returned.
Otherwise, bytes representing the artifact object will be returned.
Parameters
----------
key : str
Name of the artifact.
Returns
-------
str or object or bytes
Path of the artifact, the artifact object, or a bytestream representing the
artifact.
"""
artifact = self._get_artifact(
key, _CommonCommonService.ArtifactTypeEnum.BLOB)
artifact_stream = six.BytesIO(artifact)
torch = importer.maybe_dependency("torch")
if torch is not None:
try:
obj = torch.load(artifact_stream)
except: # not something torch can deserialize
artifact_stream.seek(0)
else:
artifact_stream.close()
return obj
try:
obj = pickle.load(artifact_stream)
except: # not something pickle can deserialize
artifact_stream.seek(0)
else:
artifact_stream.close()
return obj
return artifact_stream
# -*- coding: utf-8 -*-
import pytest
from verta import data_types
from verta._internal_utils import importer
if importer.maybe_dependency("scipy") is None:
pytest.skip("scipy is not installed", allow_module_level=True)
class TestConfusionMatrix:
def test_confusion_matrix(self):
attr = data_types.ConfusionMatrix(
value=[[1, 2, 3], [4, 5, 6], [7, 8, 9]],
labels=["a", "b", "c"],
)
d = {
"type": "verta.confusionMatrix.v1",
"confusionMatrix": {
"labels": ["a", "b", "c"],
"value": [[1, 2, 3], [4, 5, 6], [7, 8, 9]],
},
}
assert attr._as_dict() == d
assert attr == data_types._VertaDataType._from_dict(d)
def test_confusion_matrix_numpy(self):
np = pytest.importorskip("numpy")
attr = data_types.ConfusionMatrix(
value=np.arange(1, 10).reshape((3, 3)),
labels=np.array(["a", "b", "c"]),
