|Abstract:|| In the latest years, scholars started focusing on how to develop statistical tool for the analysis of population of complex data, such as sets of labelled or unlabelled graphs graphs. The present works adds to this literature by focusing on a strangely overlooked area, namely the formulation of prediction sets.
By exploiting cutting edge techniques in the realm of machine learning, we propose a forecasting method for populations of both labelled and unlabelled graphs based on Conformal Prediction, able to identify prediction regions. Our method is model-free, achieves finite-sample validity, is computationally efficient and it identifies interpretable prediction sets, in the shape of a parallelotope. To explore the features of this novel forecasting technique, a simulation study and and a real-world example are presented.|