In recent years, global warming has become more serious problem because the amount of greenhouse gas (GHG) emissions is increased. Therefore, renewable energy such as solar energy should be utilized effectively to reduce GHG emissions. However, it is difficult to balance the supply and demand for solar energy because the output depends on the weather. The on-demand cumulative control method is a method of inventory control to balance between supply and demand dynamically. The on-demand cumulative control method involves the forecasting phase of power generation and consumption in order to estimate the effective power supply. However, when they are not taken into account the precision of forecasting electricity supply and demand, the appropriate demand-to-supply management might be difficult. This study analyzes forecasting method in solar energy demand-to-supply management by comparing the exponential smoothing method and random forest is implemented under the same explanatory variables in terms of the precision of forecasting power generation and consumption. Moreover, parameters such as temperature; date; weekdays or holidays; and period of time are set as explanatory variables and the effect of setting them is evaluated. Finally, the total power purchase amount is investigated using the on-demand cumulative control method.