def automl_predict(
input: Path,
output: Path = Path("output.csv"),
model: Path = Path("automl.bin"),
ignore_cols: List[int] = typer.Option([]),
format: str = None,
):
"""
🔮 Predict with a previously trained AutoML instance.
"""
try:
dataset = _load_dataset(format, input, ignore_cols)
except ValueError as e:
logger.error(f"⚠️ Error: {str(e)}")
return
try:
with model.open("rb") as fp:
automl = AutoML.load(fp)
except TypeError as e:
logger.error(f"⚠️ Error: {str(e)}")
return
console.print(f"🔮 Predicting {len(dataset)} items with the pipeline:")
console.print(repr(automl.best_pipeline_))
X = dataset.values
y = automl.predict(X)
with output.open("wt") as fp:
df = pd.DataFrame(y, columns=["y"])
df.to_csv(fp)
console.print(f"💾 Predictions saved to [blue]{output.absolute()}[/]")
def automl_inspect(model: Path = Path("automl.bin")):
"""
🔍 Inspect a trained AutoML model.
"""
with model.open("rb") as fp:
automl = AutoML.load(fp)
console.print(f"🔍 Inspecting AutoML model: [green]{model.absolute()}[/]")
console.print(f"⭐ Best pipeline (score={automl.best_score_:0.3f}):")
console.print(repr(automl.best_pipeline_))
def test_automl_save_load():
X, y = dummy.generate(seed=0)
automl = AutoML(search_iterations=3, registry=[DummyAlgorithm])
automl.fit(X, y)
pipe = automl.best_pipeline_
fp = BytesIO()
automl.save(fp)
fp.seek(0)
automl2 = AutoML.load(fp)
pipe2 = automl2.best_pipeline_
assert repr(pipe) == repr(pipe2)
def test_automl_save_load():
X, y = dummy.generate(seed=0)
automl = AutoML(
input=(MatrixContinuousDense, Supervised[VectorCategorical]),
output=VectorCategorical,
search_iterations=3,
registry=[DummyAlgorithm],
)
automl.fit(X, y)
pipe = automl.best_pipeline_
fp = BytesIO()
automl.save(fp)
fp.seek(0)
automl2 = AutoML.load(fp)
pipe2 = automl2.best_pipeline_
assert repr(pipe) == repr(pipe2)