try:
                top_k_vars=SelectKBest(chi2, k=self.top_k)
                top_k_vars.fit_transform(self.data[columns], self.data[self.target])