def score_from_file(
self,
file_path: str,
target_column: Optional[str] = None,
encoding: Optional[str] = None,
**kwargs,
) -> float:
""" Calculate `self.scoring` metric of the model on data in the file.
Parameters
----------
file_path: str
A csv or ARFF file with which to calculate the score.
target_column: str, optional (default=None)
Specifies which column the model should predict.
If left None, the last column is taken to be the target.
encoding: str, optional
Encoding of the ARFF file.
**kwargs:
Any additional arguments for calls to pandas.read_csv or arff.load.
Returns
-------
float
The score obtained on the given test data according to the `scoring` metric.
"""
x, y = X_y_from_file(file_path,
split_column=target_column,
encoding=encoding,
**kwargs)
return self.score(x, y)
def fit_from_file(
self,
file_path: str,
target_column: Optional[str] = None,
encoding: Optional[str] = None,
warm_start: Optional[List[Individual]] = None,
**kwargs,
) -> None:
""" Find and fit a model to predict the target column (last) from other columns.
Parameters
----------
file_path: str
Path to a csv or ARFF file containing the training data.
target_column: str, optional (default=None)
Specifies which column the model should predict.
If left None, the last column is taken to be the target.
encoding: str, optional
Encoding of the file.
warm_start: List[Individual], optional (default=None)
A list of individual to start the search procedure with.
If None is given, random start candidates are generated.
**kwargs:
Any additional arguments for calls to pandas.read_csv or arff.load.
"""
x, y = X_y_from_file(file_path, target_column, encoding, **kwargs)
self.fit(x, y, warm_start)
def predict_proba_from_file(
self,
arff_file_path: str,
target_column: Optional[str] = None,
encoding: Optional[str] = None,
):
""" Predict the class probabilities for input in the arff_file.
Parameters
----------
arff_file_path: str
An ARFF file with the same columns as the one that used in fit.
Target column must be present in file, but its values are ignored.
target_column: str, optional (default=None)
Specifies which column the model should predict.
If left None, the last column is taken to be the target.
encoding: str, optional
Encoding of the ARFF file.
Returns
-------
numpy.ndarray
Numpy array with class probabilities.
The array is of shape (N, K) where N is len(X),
and K is the number of class labels found in `y` of `fit`.
"""
x, _ = X_y_from_file(arff_file_path, target_column, encoding)
x = self._prepare_for_prediction(x)
return self._predict_proba(x)
def predict_from_file(
self,
file_path: str,
target_column: Optional[str] = None,
encoding: Optional[str] = None,
**kwargs,
) -> np.ndarray:
""" Predict the target for input found in the ARFF file.
Parameters
----------
file_path: str
A csv or ARFF file with the same columns as the one that used in fit.
Target column must be present in file, but its values are ignored.
target_column: str, optional (default=None)
Specifies which column the model should predict.
If left None, the last column is taken to be the target.
encoding: str, optional
Encoding of the ARFF file.
**kwargs:
Any additional arguments for calls to pandas.read_csv or arff.load.
Returns
-------
numpy.ndarray
array with predictions for each row in the ARFF file.
"""
x, _ = X_y_from_file(file_path,
split_column=target_column,
encoding=encoding,
**kwargs)
x = self._prepare_for_prediction(x)
return self._predict(x)
def main():
args = parse_args()
print("CLI: Processing input")
if not os.path.exists(args.input_file.lower()):
raise FileNotFoundError(args.input_file)
if args.input_file.lower().split(".")[-1] not in ["csv", "arff"]:
raise ValueError("Unknown file extension. Please use csv or arff.")
kwargs = {}
if args.input_file.lower().endswith(".csv") and args.separator is not None:
kwargs["sep"] = args.seperator
x, y = X_y_from_file(
file_path=args.input_file.lower(),
split_column=args.target,
**kwargs,
)
if args.mode is None:
if is_categorical_dtype(y.dtype):
args.mode = "classification"
else:
args.mode = "regression"
print(f"Detected a {args.mode} problem.")
print("CLI: Initializing GAMA")
log_level = logging.INFO if args.verbose else logging.WARNING
configuration = dict(
regularize_length=args.prefer_short,
max_total_time=args.time_limit_m * 60,
max_eval_time=args.max_eval_time_m * 60,
n_jobs=args.n_jobs,
verbosity=log_level,
output_directory=args.outdir,
store="nothing" if args.dry_run else "logs",
)
if args.metric:
configuration["scoring"] = args.metric
if args.mode == "regression":
automl = GamaRegressor(**configuration)
elif args.mode == "classification":
automl = GamaClassifier(**configuration)
else:
raise ValueError(f"Mode {args.mode} is not valid (--mode).")
if not args.dry_run:
print("CLI: Starting model search")
automl.fit(x, y)
# == Model Export ===
print("CLI: Exporting models.")
with open(args.output_file, "wb") as fh:
pickle.dump(automl.model, fh)
if args.export_python is not None:
automl.export_script(args.export_python, raise_if_exists=False)
else:
automl.cleanup("all")
print("done!")
def test_X_y_from_file_default_split_column(self):
_, y = X_y_from_file(ARFF_CJS)
assert y.name == "INTERNODE_29"
def test_X_y_from_file_invalid_split_column(self):
with pytest.raises(ValueError,
match="No column named NOT_EXIST found"):
X_y_from_file(ARFF_CJS, split_column="NOT_EXIST")
def test_X_y_from_arff(self):
x, y = X_y_from_file(ARFF_CJS, split_column="TR")
_test_x_y_d23380(x, y)
def test_X_y_from_csv(self):
x, y = X_y_from_file(CSV_CJS_FULL, split_column="TR")
_test_x_y_d23380(x, y)