`efc`.EnergyBasedFlowClassifier¶

class efc.EnergyBasedFlowClassifier(pseudocounts=0.5, cutoff_quantile=0.95, n_bins=30, n_jobs=None)[source]¶

The Energy-based Flow Classifier algorithm.

Parameters:

pseudocountsfloat, default=`0.5`: The weight of the pseudocounts added to empirical frequencies. Must be in the interval (0,1).
cutoff_quantilefloat, default=`0.95`: The quantile used to define the model’s energy threshold. It must be in range (0,1).
n_binsint, default=`30`: The number of bins to produce when discretizing data features. Using the quantile strategy.
n_jobsint, default=None: The number of parallel jobs to run on fit() and predict(). None means 1 unless in a joblib.parallel_backend context. -1 means using all processors.

Attributes:

max_bin_int: The maximum value of the features in X.
n_features_in_int: The number of features in X.
classes_ndarray, shape (n_classes,): The classes seen at fit().
target_type_string: The type of target seen at fit() according to utils.multiclass.type_of_target().
base_class_idx_int: The index of the base class passed to fit() in the classes_ vector. Only used when target is binary.
estimators_list of BaseEFC instances: The collection of fitted sub-estimators. When the target is binary, this collection consists of only one estimator.

__init__(pseudocounts=0.5, cutoff_quantile=0.95, n_bins=30, n_jobs=None)[source]¶

fit(X, y, base_class=None, categorical_columns=[])[source]¶

Fit the Energy-based Flow Classifier model according to X.

Parameters:

Xarray-like, shape (n_samples, n_features): The training input samples.
yarray-like, shape (n_samples,): The target values.
base_classint or string, depending on y’s dtype: Only used for binary target. Defines the class that will be used for training among the classes in the target vector. If no class is passed, the first class in the array np.unique(y) will be used.
categorical_columnsarray-like: Indicates categorical attributes so that they are not normalized and discretized as numeric attributes. These attributes must be encoded before being passed to EFC.

Returns:

predict(X, return_energies=False, unknown_class=False)[source]¶

Perform classification on samples in X.

Parameters:

Xarray-like, shape (n_samples, n_features): Input samples for classification.
return_energiesboolean, default=False,: Whether to return the energy vector of samples in X.
unknown_classboolean, default=False,: Whether to use the unknown class for samples with low similarity to all training classes. If targets dtype is numeric, the unknown class will be represented by -1.

Returns:

y_predarray-like, shape (n_samples, ): Class labels for samples in X.
y_energiesarray-like, shape (n_samples, ): Computed energies for samples in X.

Examples using `efc.EnergyBasedFlowClassifier`¶

Understanding EFC's predictions