Classification

class rulekit.classification.RuleClassifier(minsupp_new: float = 0.05, induction_measure: Measures = Measures.Correlation, pruning_measure: Measures | str = Measures.Correlation, voting_measure: Measures = Measures.Correlation, max_growing: float = 0.0, enable_pruning: bool = True, ignore_missing: bool = False, max_uncovered_fraction: float = 0.0, select_best_candidate: bool = False, complementary_conditions: bool = False, control_apriori_precision: bool = True, max_rule_count: int = 0, approximate_induction: bool = False, approximate_bins_count: int = 100)

Classification model.

Parameters:

  • minsupp_new (float = 5.0) – a minimum number (or fraction, if value < 1.0) of previously uncovered examples to be covered by a new rule (positive examples for classification problems); default: 5,

  • induction_measure (rulekit.params.Measures = rulekit.params.            Measures.Correlation) – measure used during induction; default measure is correlation

  • pruning_measure (Union[rulekit.params.Measures, str] = rulekit.params.Measures.Correlation) –

    measure used during pruning. Could be user defined (string), for example

    2 * p / n; default measure is correlation

  • voting_measure (rulekit.params.Measures = rulekit.params.Measures.Correlation) – measure used during voting; default measure is correlation

  • max_growing (int = 0.0) – non-negative integer representing maximum number of conditions which can be added to the rule in the growing phase (use this parameter for large datasets if execution time is prohibitive); 0 indicates no limit; default: 0

  • enable_pruning (bool = True) – enable or disable pruning, default is True.

  • ignore_missing (bool = False) – boolean telling whether missing values should be ignored (by default, a missing valueof given attribute is always cconsidered as not fulfilling the condition build upon that attribute); default: False.

  • max_uncovered_fraction (float = 0.0) – Floating-point number from [0,1] interval representing maximum fraction of examples that may remain uncovered by the rule set, default: 0.0.

  • select_best_candidate (bool = False) – Flag determining if best candidate should be selected from growing phase; default: False.

  • complementary_conditions (bool = False) – If enabled, complementary conditions in the form a = !{value} for nominal attributes are supported.

  • control_apriori_precision (bool = True) – When inducing classification rules, verify if candidate precision is higher than apriori precision of the investigated class.

  • max_rule_count (int = 0) – Maximum number of rules to be generated (for classification data sets it applies to a single class); 0 indicates no limit.

  • approximate_induction (bool = False) – Use an approximate induction heuristic which does not check all possible splits; note: this is an experimental feature and currently works only for classification data sets, results may change in future;

  • approximate_bins_count (int = 100) – maximum number of bins for an attribute evaluated in the approximate induction.

add_event_listener(listener: RuleInductionProgressListener)
Add event listener object to the operator which allows to monitor

rule induction progress.

Example

>>> from rulekit.events import RuleInductionProgressListener
>>> from rulekit.classification import RuleClassifier
>>>
>>> class MyEventListener(RuleInductionProgressListener):
>>>     def on_new_rule(self, rule):
>>>         print('Do something with new rule', rule)
>>>
>>> operator = RuleClassifier()
>>> operator.add_event_listener(MyEventListener())
Parameters:

listener (RuleInductionProgressListener) – listener object

fit(values: ndarray | DataFrame | list, labels: ndarray | DataFrame | list) → RuleClassifier

Train model on given dataset.

Parameters:
Returns:

self

Return type:

RuleClassifier

get_coverage_matrix(values: ndarray | DataFrame | list) → ndarray

Calculates coverage matrix for ruleset.

Parameters:

values (rulekit.operator.Data) – dataset

Returns:

coverage_matrix – Each row of the matrix represent single example from dataset and every column represent on rule from rule set. Value 1 in the matrix cell means that rule covered certain example, value 0 means that it doesn’t.

Return type:

np.ndarray

get_metadata_routing() → None

Warning

Scikit-learn metadata routing is not supported yet.

Raises:

NotImplementedError – _description_

get_params(deep: bool = True) → dict[str, Any]
Parameters:

deep (rulekit.operator.Data) – Parameter for scikit-learn compatibility. Not used.

Returns:

hyperparameters – Dictionary containing model hyperparameters.

Return type:

np.ndarray

predict(values: ndarray | DataFrame | list, return_metrics: bool = False) → ndarray | tuple[ndarray, ClassificationPredictionMetrics]

Perform prediction and returns predicted labels.

Parameters:

  • values (rulekit.operator.Data) – attributes

  • return_metrics (bool = False) – Optional flag. If set to True method will calculate some additional model metrics. Method will then return tuple instead of just predicted labels.

Returns:

result – If return_metrics flag wasn’t set it will return just prediction, otherwise a tuple will be returned with first element being prediction and second one being metrics.

Return type:

Union[np.ndarray, tuple[np.ndarray, rulekit.classification.            ClassificationPredictionMetrics]]

predict_proba(values: ndarray | DataFrame | list, return_metrics: bool = False) → ndarray | tuple[ndarray, ClassificationPredictionMetrics]

Perform prediction and returns class probabilities for each example.

Parameters:

  • values (rulekit.operator.Data) – attributes

  • return_metrics (bool = False) – Optional flag. If set to True method will calculate some additional model metrics. Method will then return tuple instead of just probabilities.

Returns:

result – If return_metrics flag wasn’t set it will return just probabilities matrix, otherwise a tuple will be returned with first element being prediction and second one being metrics.

Return type:

Union[np.ndarray, tuple[np.ndarray, rulekit.classification.            ClassificationPredictionMetrics]]

score(values: ndarray | DataFrame | list, labels: ndarray | DataFrame | list) → float

Return the accuracy on the given test data and labels.

Parameters:
Returns:

score – Accuracy of self.predict(values) wrt. labels.

Return type:

float

set_params(**kwargs) → object

Set models hyperparameters. Parameters are the same as in constructor.

class rulekit.classification.ExpertRuleClassifier(minsupp_new: float = 0.05, induction_measure: Measures = Measures.Correlation, pruning_measure: Measures | str = Measures.Correlation, voting_measure: Measures = Measures.Correlation, max_growing: float = 0.0, enable_pruning: bool = True, ignore_missing: bool = False, max_uncovered_fraction: float = 0.0, select_best_candidate: bool = False, complementary_conditions: bool = False, control_apriori_precision: bool = True, max_rule_count: int = 0, approximate_induction: bool = False, approximate_bins_count: int = 100, extend_using_preferred: bool = False, extend_using_automatic: bool = False, induce_using_preferred: bool = False, induce_using_automatic: bool = False, consider_other_classes: bool = False, preferred_conditions_per_rule: int = 2147483647, preferred_attributes_per_rule: int = 2147483647)

Classification model using expert knowledge.

Parameters:

  • minsupp_new (float = 5.0) –

  • fraction (a minimum number (or) –

  • examples (if value < 1.0) of previously uncovered) –

  • problems); (to be covered by a new rule (positive examples for classification) –

  • default (5,) –

  • induction_measure (rulekit.params.Measures = rulekit.params.Measures.Correlation) – measure used during induction; default measure is correlation

  • pruning_measure (Union[rulekit.params.Measures, str] = rulekit.params.Measures.Correlation) –

    measure used during pruning. Could be user defined (string), for example

    2 * p / n; default measure is correlation

  • voting_measure (rulekit.params.Measures = rulekit.params.Measures.Correlation) – measure used during voting; default measure is correlation

  • max_growing (int = 0.0) – non-negative integer representing maximum number of conditions which can be added to the rule in the growing phase (use this parameter for large datasets if execution time is prohibitive); 0 indicates no limit; default: 0

  • enable_pruning (bool = True) – enable or disable pruning, default is True.

  • ignore_missing (bool = False) – boolean telling whether missing values should be ignored (by default, a missing value of given attribute is always considered as not fulfilling the condition build upon that attribute); default: False.

  • max_uncovered_fraction (float = 0.0) – Floating-point number from [0,1] interval representing maximum fraction of examples that may remain uncovered by the rule set, default: 0.0.

  • select_best_candidate (bool = False) – Flag determining if best candidate should be selected from growing phase; default: False.

  • complementary_conditions (bool = False) – If enabled, complementary conditions in the form a = !{value} for nominal attributes are supported.

  • control_apriori_precision (bool = True) – When inducing classification rules, verify if candidate precision is higher than apriori precision of the investigated class.

  • max_rule_count (int = 0) – Maximum number of rules to be generated (for classification data sets it applies to a single class); 0 indicates no limit.

  • approximate_induction (bool = False) – Use an approximate induction heuristic which does not check all possible splits; note: this is an experimental feature and currently works only for classification data sets, results may change in future;

  • approximate_bins_count (int = 100) – maximum number of bins for an attribute evaluated in the approximate induction.

  • extend_using_preferred (bool = False) – boolean indicating whether initial rules should be extended with a use of preferred conditions and attributes; default is False

  • extend_using_automatic (bool = False) – boolean indicating whether initial rules should be extended with a use of automatic conditions and attributes; default is False

  • induce_using_preferred (bool = False) – boolean indicating whether new rules should be induced with a use of preferred conditions and attributes; default is False

  • induce_using_automatic (bool = False) – boolean indicating whether new rules should be induced with a use of automatic conditions and attributes; default is False

  • consider_other_classes (bool = False) – boolean indicating whether automatic induction should be performed for classes for which no user’s knowledge has been defined (classification only); default is False.

  • preferred_conditions_per_rule (int = None) – maximum number of preferred conditions per rule; default: unlimited,

  • preferred_attributes_per_rule (int = None) – maximum number of preferred attributes per rule; default: unlimited.

add_event_listener(listener: RuleInductionProgressListener)
Add event listener object to the operator which allows to monitor

rule induction progress.

Example

>>> from rulekit.events import RuleInductionProgressListener
>>> from rulekit.classification import RuleClassifier
>>>
>>> class MyEventListener(RuleInductionProgressListener):
>>>     def on_new_rule(self, rule):
>>>         print('Do something with new rule', rule)
>>>
>>> operator = RuleClassifier()
>>> operator.add_event_listener(MyEventListener())
Parameters:

listener (RuleInductionProgressListener) – listener object

fit(values: ndarray | DataFrame | list, labels: ndarray | DataFrame | list, expert_rules: list[str | tuple[str, str]] | None = None, expert_preferred_conditions: list[str | tuple[str, str]] | None = None, expert_forbidden_conditions: list[str | tuple[str, str]] | None = None) → ExpertRuleClassifier

Train model on given dataset.

Parameters:

  • values (rulekit.operator.Data) – attributes

  • labels (rulekit.operator.Data) – labels

  • expert_rules (List[Union[str, Tuple[str, str]]]) – set of initial rules, either passed as a list of strings representing rules or as list of tuples where first element is name of the rule and second one is rule string.

  • expert_preferred_conditions (List[Union[str, Tuple[str, str]]]) – multiset of preferred conditions (used also for specifying preferred attributes by using special value Any). Either passed as a list of strings representing rules or as list of tuples where first element is name of the rule and second one is rule string.

  • expert_forbidden_conditions (List[Union[str, Tuple[str, str]]]) – set of forbidden conditions (used also for specifying forbidden attributes by using special value Any). Either passed as a list of strings representing rules or as list of tuples where first element is name of the rule and second one is rule string.

Returns:

self

Return type:

ExpertRuleClassifier

get_coverage_matrix(values: ndarray | DataFrame | list) → ndarray

Calculates coverage matrix for ruleset.

Parameters:

values (rulekit.operator.Data) – dataset

Returns:

coverage_matrix – Each row of the matrix represent single example from dataset and every column represent on rule from rule set. Value 1 in the matrix cell means that rule covered certain example, value 0 means that it doesn’t.

Return type:

np.ndarray

get_metadata_routing() → None

Warning

Scikit-learn metadata routing is not supported yet.

Raises:

NotImplementedError – _description_

get_params(deep: bool = True) → dict[str, Any]
Parameters:

deep (rulekit.operator.Data) – Parameter for scikit-learn compatibility. Not used.

Returns:

hyperparameters – Dictionary containing model hyperparameters.

Return type:

np.ndarray

predict(values: ndarray | DataFrame | list, return_metrics: bool = False) → ndarray | tuple[ndarray, ClassificationPredictionMetrics]

Perform prediction and returns predicted labels.

Parameters:

  • values (rulekit.operator.Data) – attributes

  • return_metrics (bool = False) – Optional flag. If set to True method will calculate some additional model metrics. Method will then return tuple instead of just predicted labels.

Returns:

result – If return_metrics flag wasn’t set it will return just prediction, otherwise a tuple will be returned with first element being prediction and second one being metrics.

Return type:

Union[np.ndarray, tuple[np.ndarray, rulekit.classification.            ClassificationPredictionMetrics]]

predict_proba(values: ndarray | DataFrame | list, return_metrics: bool = False) → ndarray | tuple[ndarray, ClassificationPredictionMetrics]

Perform prediction and returns class probabilities for each example.

Parameters:

  • values (rulekit.operator.Data) – attributes

  • return_metrics (bool = False) – Optional flag. If set to True method will calculate some additional model metrics. Method will then return tuple instead of just probabilities.

Returns:

result – If return_metrics flag wasn’t set it will return just probabilities matrix, otherwise a tuple will be returned with first element being prediction and second one being metrics.

Return type:

Union[np.ndarray, tuple[np.ndarray, rulekit.classification.            ClassificationPredictionMetrics]]

score(values: ndarray | DataFrame | list, labels: ndarray | DataFrame | list) → float

Return the accuracy on the given test data and labels.

Parameters:
Returns:

score – Accuracy of self.predict(values) wrt. labels.

Return type:

float

set_params(**kwargs) → object

Set models hyperparameters. Parameters are the same as in constructor.