Nowadays, renewable energy forecasting has become a crucial tool for power system utilities to mitigate the complexities of integrating variable renewables with the grid. Univariate approach of forecasting is gradually growing for renewable energy in order to reduce the computational efforts and dependence on weather data. However, the initial process of selecting an appropriate forecasting model for different renewables in univariate forecasting is an ambiguous task. Therefore, this chapter explores major research problems encountered during this process through an empirical analysis of different statistical and machine learning models employed for univariate forecasting of different renewables (hydro and wind). Most widely used forecasting models, viz. ETS, SARIMA, SVR, and LSTM, are utilized to examine the forecasting performance of machine learning and statistical models over different forecasting horizons. Findings of this study show that statistical models can outperform deep learning and machine learning models in univariate time-series forecasting of renewable energy. Results demonstrate that ETS and SARIMA exhibit better accuracy for wind and hydro energy, respectively, over shorter as well as longer forecasting periods.