def meta_val(x_pred, y_real, x_exp, data, x_prob, a, setting, lower_bound, upper_bound): if len(a) == 0: setting['exp_x'] = x_exp setting['exp_y'] = data[:, 2] setting['bounds'] = [(lower_bound, upper_bound), (lower_bound, upper_bound), (lower_bound, upper_bound)] model = createKriging.createKriging(setting) y_kriging, ci_kriging, var_kriging = createKriging.predictor( x_pred, model, setting) y_pred = y_kriging y_Real = y_real[:, 2] rmse, nrmse = utilPCE.validation(y_Real, y_pred) else: rmse = npy.zeros(len(a)) nrmse = npy.zeros(len(a)) for i in xrange(0, len(a)): setting['exp_x'] = x_exp setting['exp_y'] = data[i * len(x_exp):(i + 1) * len(x_exp), 2] setting['bounds'] = [(lower_bound, upper_bound), (lower_bound, upper_bound)] model = createKriging.createKriging(setting) y_kriging, ci_kriging, var_kriging = createKriging.predictor( x_pred, model, setting) y_pred = y_kriging y_Real = y_real[i * len(x_pred):(i + 1) * len(x_pred), 2] rmse[i], nrmse[i] = utilPCE.validation(y_Real, y_pred) return rmse, nrmse
def meta_val(n_deg, x_pred, y_real, x_exp, data, x_prob, a, meta_type): if len(a) == 0: x_experiment = x_exp y_experiment = data[:, 2] PCE = calPCE.collocation(n_deg, x_prob, x_experiment, y_experiment, meta_type) PCE_pred, y_pred = utilPCE.predictor(PCE, x_pred) y_Real = y_real[:, 2] rmse, nrmse = utilPCE.validation(y_Real, y_pred) else: rmse = npy.zeros(len(a)) nrmse = npy.zeros(len(a)) data_meta = npy.zeros((len(a) * len(x_pred), 3)) for i in xrange(0, len(a)): x_experiment = x_exp y_experiment = data[i * len(x_exp):(i + 1) * len(x_exp), 2] PCE = calPCE.collocation(n_deg, x_prob, x_experiment, y_experiment, meta_type) PCE_pred, y_pred = utilPCE.predictor(PCE, x_pred) y_Real = y_real[i * len(x_pred):(i + 1) * len(x_pred), 2] rmse[i], nrmse[i] = utilPCE.validation(y_Real, y_pred) return rmse, nrmse