The total solar irradiance (TSI) varies on timescales of minutes to centuries. On short timescales it varies due to the superposition of intensity fluctuations produced by turbulent convection and acoustic oscillations. On longer timescales, it changes due to photospheric magnetic activity, mainly because of the facular brightenings and dimmings caused by sunspots. While modern TSI variations have been monitored from space since the 1970s, TSI variations over much longer periods can only be estimated either using historical observations of magnetic features, possibly supported by flux transport models, or from the measurements of the cosmogenic isotope concentrations in tree rings and ice cores. The reconstruction of the TSI in the last few centuries, particularly in the 17th/18th centuries during the Maunder minimum, is of primary importance for studying climatic effects. To separate the temporal components of the irradiance variations, specifically the 11-yr magnetic cycle from secular variability, we decomposed the signals associated with historical observations of magnetic features and the solar modulation potential by applying an empirical mode decomposition algorithm. Using essentially the same approach and empirical relationships of solar activity in even-odd cycles we estimated, in 2021, the area coverage of active regions and therefore the TSI for solar cycle 25.