From 362a510dcf9ef6d9af0d8fee8f4cb68b125675fa Mon Sep 17 00:00:00 2001 From: Heitor Date: Mon, 29 Apr 2024 22:11:55 +1200 Subject: [PATCH] docs(EFDT): Updating the EFDT documentation --- src/capymoa/classifier/_efdt.py | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) diff --git a/src/capymoa/classifier/_efdt.py b/src/capymoa/classifier/_efdt.py index 89cbe0c3..c693991b 100644 --- a/src/capymoa/classifier/_efdt.py +++ b/src/capymoa/classifier/_efdt.py @@ -21,7 +21,7 @@ class EFDT(MOAClassifier): Regression Trees - CART). Keep in mind that such propositions hold when processing a stationary data stream. When dealing with non-stationary data, EFDT is somewhat robust to concept drifts as it continually revisits and updates its internal - decision tree structure. Still, in such cases, the Hoeffind Adaptive Tree might + decision tree structure. Still, in such cases, the Hoeffding Adaptive Tree might be a better option, as it was specifically designed to handle non-stationarity. Reference: @@ -44,9 +44,9 @@ class EFDT(MOAClassifier): 84.39999999999999 """ - MAJORITY_CLASS = 0 - NAIVE_BAYES = 1 - NAIVE_BAYES_ADAPTIVE = 2 + # MAJORITY_CLASS = 0 + # NAIVE_BAYES = 1 + # NAIVE_BAYES_ADAPTIVE = 2 def __init__( self, @@ -57,7 +57,7 @@ def __init__( split_criterion: Union[str, SplitCriterion] = "InfoGainSplitCriterion", confidence: float = 1e-3, tie_threshold: float = 0.05, - leaf_prediction: int = MAJORITY_CLASS, + leaf_prediction: str = "MajorityClass", nb_threshold: int = 0, numeric_attribute_observer: str = "GaussianNumericAttributeClassObserver", binary_split: bool = False, @@ -77,10 +77,8 @@ def __init__( :param confidence: Significance level to calculate the Hoeffding bound. The significance level is given by `1 - delta`. Values closer to zero imply longer split decision delays. :param tie_threshold: Threshold below which a split will be forced to break ties. - :param leaf_prediction: Prediction mechanism used at leafs. - - 0 - Majority Class - - 1 - Naive Bayes - - 2 - Naive Bayes Adaptive + :param leaf_prediction: Prediction mechanism used at leafs + (0: Majority Class, 1: Naive Bayes, 2: Naive Bayes Adaptive). :param nb_threshold: Number of instances a leaf should observe before allowing Naive Bayes. :param numeric_attribute_observer: The Splitter or Attribute Observer (AO) used to monitor the class statistics of numeric features and perform splits.