Enviar Registro por Correo electrónico: Smoothing and stationarity enforcement framework for deep learning time-series forecasting