Estimation of the Ultraviolet Index Using Artificial Intelligence Techniques

Abstract

Ultraviolet radiation poses a significant risk to human health, as it is associated with sunburn, premature aging, skin cancer, and ocular damage. Peru records UV index levels ranging from 14 to 20, which are considered among the highest worldwide, thereby highlighting the need for reliable monitoring and prediction systems. In this study, a predictive model was developed to estimate the maximum UV index using meteorological variables, including temperature, humidity, heat index, barometric pressure, solar radiation, and solar energy, through the application of machine learning techniques to historical records collected from the DAVIS Vantage PRO 2.0 meteorological station located in Puno during the 2017–2024 period. Following data cleaning, exploratory analysis, and model training, linear regression and Random Forest approaches were evaluated, with solar radiation and solar energy identified as the most influential predictors. Although linear regression achieved a satisfactory fit (R2 = 0.87), the Random Forest algorithm demonstrated superior performance (R2 = 0.95, MAE ≈ 0.75), with homogeneously distributed residuals and no evidence of systematic bias. Consequently, decision tree–based methods emerge as robust tools for real-time UV index estimation, with potential applications in public health, risk prevention, and the planning of outdoor activities.