prototypes/dml_iv/dml_ate_iv.py [93:120]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        self._effect = np.mean(res_y * res_z)/np.mean(res_t * res_z)

        self._std = np.std(res_y * res_z)/(np.sqrt(res_y.shape[0]) * np.abs(np.mean(res_t * res_z)))

        return self

    def effect(self, X=None):
        """
        Parameters
        ----------
        X : features
        """
        if X is None:
            return self._effect
        else:
            return self._effect * np.ones(X.shape[0])

    def normal_effect_interval(self, lower=5, upper=95):
        return (scipy.stats.norm.ppf(lower/100, loc=self._effect, scale=self._std),
                scipy.stats.norm.ppf(upper/100, loc=self._effect, scale=self._std))
    @property
    def std(self):
        return self._std

    @property
    def fitted_nuisances(self):
        return {'model_Y_X': self.model_Y_X,
                'model_T_X': self.model_T_X,
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



prototypes/dml_iv/dml_ate_iv.py [205:233]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        self._effect = np.mean(res_y * res_z)/np.mean(res_t * res_z)

        self._std = np.std(res_y * res_z)/(np.sqrt(res_y.shape[0]) * np.abs(np.mean(res_t * res_z)))

        return self

    def effect(self, X=None):
        """
        Parameters
        ----------
        X : features
        """
        if X is None:
            return self._effect
        else:
            return self._effect * np.ones(X.shape[0])

    def normal_effect_interval(self, lower=5, upper=95):
        return (scipy.stats.norm.ppf(lower/100, loc=self._effect, scale=self._std),
                scipy.stats.norm.ppf(upper/100, loc=self._effect, scale=self._std))

    @property
    def std(self):
        return self._std

    @property
    def fitted_nuisances(self):
        return {'model_Y_X': self.model_Y_X,
                'model_T_X': self.model_T_X,
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -