Nations worldwide recognize water as an essential natural resource that plays a pivotal role in their socioeconomic development. Advances in technology and scientific research have significantly enhanced the understanding of Earth's water systems, enabling contributions to their sustainable use and management and ensuring the availability of water for future generations. However, with the impacts of climate change, agricultural intensification, industrialization, and rapid population growth, a surge in water demand has been observed, while rainfall levels remain nearly constant. This mismatch between water supply and demand has made the effective planning and management of water resources more critical than ever. To address this issue, developing and applying data-driven rainfall-runoff models are seen as a viable solution. In this study, daily and monthly rainfall-runoff predictive models were developed and tested using Support Vector Machine (SVM), Random Forest (RF), and Artificial Neural Network (ANN) techniques in the MATLAB 2019b interface for the Gumara Watershed in Lake Tana Sub-basin. Climatic data were used as independent variables, while hydrological data were treated as dependent variables, with uniform watershed characteristics being assumed. Among the models, consistent performance was exhibited by the SVM having R2 values of 0.70,0.87 and 0.70,0.85, the highest accuracy was achieved RF the with the values of R2 0.80,0.90 and 0.72, 0.85 during training and testing for daily and monthly datasets respectively, and the ANN demonstrated intermediate results. In conclusion, while further research incorporating watershed characteristics as input variables is recommended, the RF model is a promising solution for water resource planning and management in the Gumara Watershed and other data-scare areas, with benefits for the local community. Keywords: ANN, Hydrological modeling, Rainfall-Runoff modeling, RF, SVM