A pratice with the Rossmann dataset on the Kaggle competition