Dimensionality Reduction in Solar Radiation Forecasting: A Combined PCA-SVM Framework for Renewable Energy Applications

Sharief Basha Shaik^*Nagaraja Rao A

• VIT University, Vellore, India

This study presents a comprehensive framework for solar radiation forecasting (SRF) by integrating Principal Component Analysis (PCA) for dimensionality reduction with Support Vector Regression (SVR) for prediction. The research utilizes meteorological data collected throughout 2023 at VIT University's campus in Vellore, Tamil Nadu, incorporating multiple parameters including ambient temperature, dew point, wind characteristics, and atmospheric conditions. To address challenges of high-dimensional data affecting model generalization, PCA was employed to transform correlated variables into uncorrelated principal components while preserving essential patterns. The methodology involved data normalization, covariance analysis, component extraction, and selective feature retention based on cumulative explained variance thresholds. The dimensionally reduced dataset was then fed into various SVR models with different kernel functions (linear, polynomial, tanh, and Gaussian), and model validation was rigorously performed using k-fold cross-validation to identify the optimal configuration for solar radiation prediction. Comparative analysis revealed that the SVR_Gaussian implementation demonstrated superior performance with optimal values of RMSE (9.111125), MAE (3.765607), MAPE (0.013853), and R² (98.37%), outperforming all alternative models. The hybrid PCA-SVR approach effectively handles the inherent complexity of solar radiation patterns while maintaining computational efficiency.