Meteorological data of the city of N'Djamena

In order to enable an efficient and reliable integration of photovoltaic power plants into powergrids in sub-Saharan Africa in general and Chad in particular, which is committed to massively exploitingsolar energy to address its energy deficit, this work aims to forecast hourly temperature and global horizontalirradiation (GHI). In this study, we employ ensemble methods represented by random forest (RF) and extremeboosting gradient (XGB) model. Their performances are then compared with the support vector regression(SVR) model. As the initial dataset consisted of daily statistics to make hourly previsions, we introduced anoriginal approach for discretising variables taking into account the nature of the meteorological parameterswith a set of assumptions about their daily trends