In the forecasting of irregular time series using modern Machine Learning methods, some methodologies have become pretty popular, such as those implemented in the XGBoost tool. XGBoost has already shown to be efficient in many situations. However, in the case of a set of time series, for instance associated with a set of geographical points and exhibiting significant correlations, XGBoost needs an important amount of features engineering to be able to capture that supplementary and useful information. In these situations, the Graph Neural Networks (GNNs) family offers tools designed to take into account these spatial correlations automatically. In this paper, we consider the time series of the daily maximal temperatures collected at many meteorological stations in a wide area, and we try to forecast the maximal temperatures in the next few days at all those points. These are pretty irregular series, and we chose a particularly powerful algorithm belonging to the GNN class called Graph WaveNet for this task. The algorithm hasn't been applied before to this type of target. We explored its behavior with no need for any feature engineering, and we also compared it with that of XGBoost for which we should invest a significant effort to exploit the spatial aspects of data. We considered the forecasting up to 10 days, the meteorological range. Basically, Graph WaveNet behaves better for very short prediction horizons and becomes less accurate when the prediction horizon increases. XGBoost stays closer to the mean of the series when the horizon gets large (close to 10 days). The paper provides more details about the two tools' behavior.