356bc2655cd30123f5ad008b5dc66b1940391331,pgmpy/sampling/Sampling.py,BayesianModelSampling,rejection_sample,#BayesianModelSampling#Any#Any#Any#,113

Before Change

            i += len(_sampled)

        if return_type.lower() == "dataframe":
            if HAS_PANDAS:
                return pandas.DataFrame.from_records(sampled)
            else:
                warn("Pandas installation not found. Returning numpy.recarray object")
                return sampled
        else:
            return sampled

    def likelihood_weighted_sample(self, evidence=None, size=1, return_type="dataframe"):
        
        Generates weighted sample(s) from joint distribution of the bayesian
        network, that comply with the given evidence.

After Change

        if evidence is None:
            return self.forward_sample(size)
        types = [(var_name, "int") for var_name in self.topological_order]
        sampled = np.zeros(0, dtype=types).view(np.recarray)
        prob = 1
        i = 0
        while i < size:
            _size = int(((size - i) / prob) * 1.5)
            _sampled = self.forward_sample(_size, "recarray")

            for evid in evidence:
                _sampled = _sampled[_sampled[evid[0]] == evid[1]]

            prob = max(len(_sampled) / _size, 0.01)
            sampled = np.append(sampled, _sampled)[:size]

            i += len(_sampled)

        return _return_samples(return_type, sampled)

    def likelihood_weighted_sample(self, evidence=None, size=1, return_type="dataframe"):
        
        Generates weighted sample(s) from joint distribution of the bayesian

Instances