Analysis and Prediction of Cloud Cover in Tanjung Pinang Using Machine Learning Concepts as a Consideration for Solar Power Plant Development

ABSTRACT
 Analysis and Prediction of Cloud Cover in Tanjung Pinang Using Machine Learning Concepts as a Consideration for Solar Power Plant Development (supervised by Prof. Dr. Halmar Halide, M.Sc). The development of renewable energy, particularly solar power plants (PLTS), greatly depends on the intensity of solar radiation reaching the Earth’s surface. One crucial factor influencing the variability of this radiation is cloud cover. Tanjung Pinang, as a tropical region with high solar energy potential, faces challenges due to fluctuating atmospheric dynamics and variable cloud cover. This study aims to analyze the relationship between cloud cover and solar radiation and to develop a predictive model of cloud cover using various machine learning algorithms to provide a technical basis for PLTS development in Tanjung Pinang. The research utilized secondary meteorological data from the Metomanz platform for recent years. The data were classified, underwent quality control, and analyzed for correlations using a heatmap. Predictive models were constructed using several machine learning algorithms such as Gradient Boosting, Logistic Regression, Decision Tree, Naïve Bayes, and Simple Linear Regression. The models were evaluated using a confusion matrix and performance metrics including accuracy, precision, and F1 score. Results. The results revealed a strong negative correlation between cloud cover and solar radiation (-0.68). The Gradient Boosting model demonstrated the best performance with an accuracy of 91.58%, a precision of 0.90, a recall of 0.92, and an F1 score of 0.90. It outperformed other models in both accuracy and stability when classifying cloudy and clear sky conditions. The developed cloud cover prediction model, particularly the Gradient Boosting algorithm, proved capable of providing accurate estimates of sky conditions in Tanjung Pinang. This is highly beneficial as a reference in assessing the feasibility of PLTS development in the region and can support decision-making systems in the utilization of solar energy in Indonesia. Moreover, this research strengthens the relevance of integrating meteorology and artificial intelligence in supporting the transition to clean and sustainable energy.

Keywords: cloud cover, machine learning, solar energy, Gradient Boosting, Tanjung Pinang