Denoising Auto Encoder + XGBoost

#1

I am trying to integrate DAE with Xgboost, I have Successfully created both the DAE and XGB model, and I guess even integrated them, but I don’t know how should I take predictions:

scores = []
final_valid_predictions = {}
for train_idx, valid_idx in KFold().split(X, Y):
    print("\nSOS\n")
    model = XGBRegressor(
        # random_state=fold,
        tree_method='gpu_hist',
        gpu_id=0,
        predictor="gpu_predictor"
    )
    xtest = df_test.copy()

    model.fit(X[train_idx], Y[train_idx])
    preds_valid = model.predict(X[valid_idx])
    # test_preds = model.predict(X_test)
    print(X_test)
    final_test_predictions.append(test_preds[gdcls])
    final_valid_predictions.update(dict(zip(valid_ids, preds_valid)))
    rmse = mean_squared_error(Y[valid_idx], preds_valid, squared=False)
    print(fold, rmse)
    scores.append(rmse)
print(np.mean(scores))

The model is broadly based on