Module fcmpy.expert_fcm.entropy

Classes

class Entropy

Entropy of the expert inputs.

Expand source code

class Entropy(ABC):
    """
        Entropy of the expert inputs.
    """
    @abstractclassmethod
    def calculateEntropy(data: pd.DataFrame, activationParamter):
        raise NotImplementedError('calculateEntropy method is not defined')

Ancestors

• abc.ABC

Subclasses

• InformationEntropy

Static methods

def calculateEntropy(activationParamter)

Expand source code

@abstractclassmethod
def calculateEntropy(data: pd.DataFrame, activationParamter):
    raise NotImplementedError('calculateEntropy method is not defined')

class InformationEntropy

Methods for calculating the information entropy.

Expand source code

class InformationEntropy(Entropy):
    """
        Methods for calculating the information entropy.
    """
    @staticmethod
    @type_check
    def calculateEntropy(**kwargs) -> pd.DataFrame:
        """
            Calculate the information entropy of the expert ratings.

            Other Parameters
            ----------------
            **data: collections.OrderedDict
                    ordered dictionary with the expert inputs
            
            Return
            -------
            y: pandas.DataFrame,
                entropy of the concept pairs in expert ratings
        """
        data = kwargs['data']
        nExperts = len(data)
        flat_data = Transform.flatData(data=data)
        prop = {}

        for concepts in set(flat_data.index):
            activation_parameter = Transform.calculateProportions(data=flat_data, 
                                            conceptPair=concepts, nExperts=nExperts) 
            prop[concepts] = activation_parameter

        entropy_concept = {}
        for concept in prop.keys():
            p = prop[concept].values()
            res = -1*sum([i*np.log2(i) for i in p if i != 0])
            res = abs(res)
            entropy_concept[concept] = res
        
        # Prepare a formated dataframe
        entropy_concept = {k:[v] for k,v in entropy_concept.items()}
        entropy_concept = pd.DataFrame(entropy_concept).T
        entropy_concept.index.set_names(['From','To'], inplace=True)
        entropy_concept.columns = ['Entropy']
        entropy_concept = entropy_concept.sort_index(level=[0,1])

        return entropy_concept

Ancestors

• Entropy
• abc.ABC

Static methods

def calculateEntropy(**kwargs) ‑> pandas.core.frame.DataFrame

Calculate the information entropy of the expert ratings.

Other Parameters

**data : collections.OrderedDict

ordered dictionary with the expert inputs

Return

y: pandas.DataFrame, entropy of the concept pairs in expert ratings

Expand source code

@staticmethod
@type_check
def calculateEntropy(**kwargs) -> pd.DataFrame:
    """
        Calculate the information entropy of the expert ratings.

        Other Parameters
        ----------------
        **data: collections.OrderedDict
                ordered dictionary with the expert inputs
        
        Return
        -------
        y: pandas.DataFrame,
            entropy of the concept pairs in expert ratings
    """
    data = kwargs['data']
    nExperts = len(data)
    flat_data = Transform.flatData(data=data)
    prop = {}

    for concepts in set(flat_data.index):
        activation_parameter = Transform.calculateProportions(data=flat_data, 
                                        conceptPair=concepts, nExperts=nExperts) 
        prop[concepts] = activation_parameter

    entropy_concept = {}
    for concept in prop.keys():
        p = prop[concept].values()
        res = -1*sum([i*np.log2(i) for i in p if i != 0])
        res = abs(res)
        entropy_concept[concept] = res
    
    # Prepare a formated dataframe
    entropy_concept = {k:[v] for k,v in entropy_concept.items()}
    entropy_concept = pd.DataFrame(entropy_concept).T
    entropy_concept.index.set_names(['From','To'], inplace=True)
    entropy_concept.columns = ['Entropy']
    entropy_concept = entropy_concept.sort_index(level=[0,1])

    return entropy_concept