sklweka package¶

sklweka.classifiers module¶

class sklweka.classifiers.WekaEstimator(*args: Any, **kwargs: Any)¶

Bases: sklearn.base.BaseEstimator, weka.core.classes.OptionHandler, sklearn.base.RegressorMixin, sklearn.base.ClassifierMixin

Wraps a Weka classifier (classifier/regressor) within the scikit-learn framework.

property classifier¶

Returns the underlying classifier object, if any.

Returns: the classifier object
Return type: Classifier

fit(X, y, sample_weight=None)¶

Trains the estimator.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – the class attribute column, array-like of shape (n_samples,)
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights. If None, then samples are equally weighted. TODO Currently ignored.

Returns

itself

Return type

WekaEstimator

get_params(deep=True)¶

Returns the parameters for this classifier, basically classname and options list.

Parameters: deep (bool) – ignored
Returns: the dictionary with options
Return type: dict

property header¶

Returns the underlying dataset header, if any.

Returns: the dataset structure
Return type: Instances

predict(X)¶

Performs predictions with the trained classifier.

Parameters: X (ndarray) – the data matrix to generate predictions for, array-like of shape (n_samples, n_features)
Returns: the score (or scores)
Return type: ndarray

predict_proba(X)¶

Performs predictions and returns class probabilities.

Parameters: X (ndarray) – the data matrix to generate predictions for, array-like of shape (n_samples, n_features)
Returns: the probabilities

score(X, y, sample_weight=None)¶

Classification: return the mean accuracy on the given test data and labels. Regression: return the coefficient of determination of the prediction.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – the class attribute column, array-like of shape (n_samples,)
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights. If None, then samples are equally weighted.

Returns

the score

Return type

float

set_params(**params)¶

Sets the options for the classifier, expects ‘classname’ and ‘options’.

Parameters: params (dict) – the parameter dictionary

sklweka.clusters module¶

class sklweka.clusters.WekaCluster(jobject=None, cluster=None, classname=None, options=None, nominal_input_vars=None, num_nominal_input_labels=None)¶

Bases: sklearn.base.BaseEstimator, weka.core.classes.OptionHandler, sklearn.base.ClusterMixin

Wraps a Weka cluster within the scikit-learn framework.

property cluster¶

Returns the underlying cluster object, if any.

Returns: the cluster object
Return type: Clusterer

fit(X, y=None, sample_weight=None)¶

Trains the cluster.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – ignored
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights. If None, then samples are equally weighted. TODO Currently ignored.

Returns

the cluster

Return type

WekaCluster

fit_predict(X, y=None, sample_weight=None)¶

Trains the cluster and returns the cluster labels.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – ignored
sample_weight (array-like of shape (n_samples,), default=None) – Sample weights. If None, then samples are equally weighted. TODO Currently ignored.

Returns

the cluster labels (of type int)

Return type

ndarray

get_params(deep=True)¶

Returns the parameters for this cluster, basically classname and options list.

Parameters: deep (bool) – ignored
Returns: the dictionary with options
Return type: dict

property header¶

Returns the underlying dataset header, if any.

Returns: the dataset structure
Return type: Instances

predict(X, y=None)¶

Predicts cluster labels.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – ignored

Returns

the cluster labels (of type int)

Return type

ndarray

set_params(**params)¶

Sets the options for the cluster, expects ‘classname’ and ‘options’.

Parameters: params (dict) – the parameter dictionary

sklweka.dataset module¶

sklweka.dataset.determine_attribute_type(y)¶

Determines the type of the column.

Parameters: y (ndarray) – the 1D vector to determine the type for
Returns: the type (C=categorical, N=numeric)
Return type: str

sklweka.dataset.determine_attribute_types(X)¶

Determines the type of the columns.

Parameters: X (ndarray) – the 2D data to determine the column types for
Returns: the list of types (C=categorical, N=numeric)
Return type: list

sklweka.dataset.load_arff(fname, class_index=None)¶

Loads the specified ARFF file. If a class index is provided, either 0-based int or 1-based string (first,second,last,last-1 are accepted as well), then the data is split into input variables and class attribute.

Parameters

fname (str) – the path of the ARFF file to load
class_index (int or str) – the class index, either int or str

Returns

tuple (X, meta), or in case of a valid class index a tuple (X,y,meta)

Return type

tuple

sklweka.dataset.load_dataset(fname, loader=None, class_index=None, internal=False)¶

Loads the dataset using Weka’s converters. If no loader instance is provided, the extension of the file is used to determine a loader (using default options). The data can either be returned using mixed types or just numeric (using Weka’s internal representation).

Parameters

fname (str) – the path of the dataset to load
loader (Loader) – the customized Loader instance to use for loading the dataset, can be None
class_index (str) – the class index string to use (‘first’, ‘second’, ‘third’, ‘last-2’, ‘last-1’, ‘last’ or 1-based index)
internal (bool) – whether to return Weka’s internal format or mixed data types

Returns

the dataset tuple: (X) if no class index; (X,y) if class index

sklweka.dataset.parse_range(r, max_value, ordered=True, safe=True)¶

Parses the Weka range string (eg “first-last” or “1,3-5,7,10-last”) of 1-based indices and returns a list of 0-based integers. ‘first’ and ‘last’ are accepted apart from integer strings, ‘-’ is used to define a range (low to high). The list can be returned ordered or as is.

Parameters

r (str) – the range string to parse
max_value (int) – the maximum value for the 1-based indices
ordered (bool) – whether to return the list ordered or as is
safe (bool) – whether to catch exceptions or not

Returns

the list of 0-base indices

Return type

list

sklweka.dataset.split_off_class(data, class_index)¶

Splits off the class attribute from the data matrix. The class index can either be a 0-based int or a 1-based string (first,second,last,last-1 are accepted as well).

Parameters

data (ndarray) – the 2D matrix to process
class_index (int or str) – the position of the class attribute to split off

Returns

the input variables (2D matrix) and the output variable (1D)

sklweka.dataset.to_array(data)¶

Turns the Instances object into ndarrays for X and y. If no class is present, then y will be None.

Parameters: data (Instances) – the data to convert
Returns: the generated arrays for X and y
Return type: tuple

sklweka.dataset.to_instance(header, x, y=None, weight=1.0)¶

Generates an Instance from the data.

Parameters

header (Instances) – the data structure to adhere to
x (ndarray) – the 1D vector with input variables
y (object) – the optional class value
weight (float) – the weight for the Instance

Returns

the generate Instance

Return type

Instance

sklweka.dataset.to_instances(X, y=None, att_names=None, att_types=None, class_name=None, class_type=None, relation_name=None, num_nominal_labels=None, num_class_labels=None)¶

Turns the 2D matrix and the optional 1D class vector into an Instances object.

Parameters

X (ndarray) – the input variables, 2D matrix
y (ndarray) – the optional class value column, 1D vector
att_names (list) – the list of attribute names
att_types – the list of attribute types (C=categorical, N=numeric), assumes numeric by default if not provided
class_name (str) – the name of the class attribute
class_type (str) – the type of the class attribute (C=categorical, N=numeric)
relation_name (str) – the name for the dataset
num_nominal_labels (dict) – the dictionary with the number of labels (key is 0-based attribute index)
num_class_labels (int) – the number of labels in the class attribute

Returns

the generated Instances object

Return type

Instances

sklweka.dataset.to_nominal_attributes(X, indices)¶

Turns the specified indices numeric column vector into a string vector.

Parameters

X (ndarray) – the 2D matrix to convert
indices (list or str) – the list of 0-based indices of attributes to convert to nominal or range string with 1-based indices

Returns

the converted matrix

Return type

ndarray

sklweka.dataset.to_nominal_labels(y)¶

Turns the numeric column vector into a string vector.

Parameters: y (list or ndarray) – the vector to convert
Returns: the converted vector
Return type: ndarray

sklweka.preprocessing module¶

class sklweka.preprocessing.MakeNominal(*args: Any, **kwargs: Any)¶

Bases: sklearn.base.BaseEstimator, sklearn.base.TransformerMixin

Converts numeric columns to nominal ones (ie string labels).

fit(X, y)¶

Trains the estimator.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – the optional class attribute column, array-like of shape (n_samples,)

Returns

itself

Return type

WekaTransformer

get_params(deep=True)¶

Returns the parameters for this classifier, basically classname and options list.

Parameters: deep (bool) – ignored
Returns: the dictionary with options
Return type: dict

property input_vars¶

Returns the 0-based indices or range string with 1-based indices of the input variables to convert.

Returns: the indices or range string, can be None
Return type: list or str

property output_vars¶

Returns whether the output variable gets converted as well.

Returns: True if the output variable gets converted
Return type: bool

set_params(**params)¶

Sets the options for the classifier, expects ‘classname’ and ‘options’.

Parameters: params (dict) – the parameter dictionary

transform(X, y=None)¶

Filters the data.

Parameters

X (ndarray) – the data to filter, array-like of shape (n_samples, n_features)
y (ndarray) – the optional class attribute column, array-like of shape (n_samples,)

Returns

the filtered data, X if no targets or (X, y) if targets provided

Return type

ndarray or tuple

class sklweka.preprocessing.WekaTransformer(*args: Any, **kwargs: Any)¶

Bases: sklearn.base.BaseEstimator, weka.core.classes.OptionHandler, sklearn.base.TransformerMixin

Wraps a Weka filter within the scikit-learn framework.

property filter¶

Returns the underlying filter object, if any.

Returns: the classifier object
Return type: Classifier

fit(X, y)¶

Trains the estimator.

Parameters

X (ndarray) – the input variables as matrix, array-like of shape (n_samples, n_features)
y (ndarray) – the optional class attribute column, array-like of shape (n_samples,)

Returns

itself

Return type

WekaTransformer

get_params(deep=True)¶

Returns the parameters for this classifier, basically classname and options list.

Parameters: deep (bool) – ignored
Returns: the dictionary with options
Return type: dict

property header¶

Returns the underlying dataset header, if any.

Returns: the dataset structure
Return type: Instances

set_params(**params)¶

Sets the options for the classifier, expects ‘classname’ and ‘options’.

Parameters: params (dict) – the parameter dictionary

transform(X, y=None)¶

Filters the data.

Parameters

X (ndarray) – the data to filter, array-like of shape (n_samples, n_features)
y (ndarray) – the optional class attribute column, array-like of shape (n_samples,)

Returns

the filtered data, X if no targets or (X, y) if targets provided

Return type

ndarray or tuple

sklweka package¶

sklweka.classifiers module¶

sklweka.clusters module¶

sklweka.dataset module¶

sklweka.preprocessing module¶

sklearn-weka-plugin

Navigation

Related Topics