ca337ebba627326cd7b15b454053e31f51b5e441,pgmpy/inference/Sampling.py,BayesianModelSampling,rejection_sample,#BayesianModelSampling#,75

Before Change

        while len(sampled) < size:
            _size = int((size - len(sampled)) / prob)
            _sampled = self.forward_sample(_size)
            for i in range(_size):
                for var in evidence:
                    if evidence[var] != _sampled[var][i]:
                        _sampled.drop(i, inplace=True)
                        break
            prob = max(len(_sampled) / _size, 0.01)  // 0.01 assumed if len(sampled) is small or zero
            sampled = sampled.append(_sampled)
        sampled.reset_index(inplace=True, drop=True)
        return sampled[:size]

After Change

            _size = int(((size - len(sampled)) / prob) * 1.5)
            _sampled = self.forward_sample(_size)
            for evid in evidence:
                _sampled = _sampled[_sampled.ix[:, evid.var] == evid]
            prob = max(len(_sampled) / _size, 0.01)
            sampled = sampled.append(_sampled)
        sampled.reset_index(inplace=True, drop=True)

Instances