Wind energy plays a major role in the global shift to sustainable and clean energy. Predicting wind speeds accurately is essential for preserving grid stability, cutting down on operating costs, and enhancing wind energy systems' overall performance. This work presents an adaptive machine learning method that uses functional data from past weather patterns to forecast wind speeds. Key meteorological variables including temperature, humidity, air pressure, dew point, and time-based characteristics are included in the dataset, which was obtained from the Open-Meteo weather API and covers the years 2024–2025.

Extensive preparation procedures were used to enhance data quality and model efficacy, including feature scaling, correlation analysis, and outlier treatment. Further aiding in the comprehension of data distributions and linkages was thorough exploratory data analysis. Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R²) were among the common performance measures used to train and assess multiple regression algorithms, such as Linear Regression, Random Forest, XGBoost, and LightGBM. The intricate, non-linear behavior of wind speeds was best modeled by ensemble-based models out of all of these.

All things considered, the results highlight how well machine learning approaches work to provide precise, real-time wind speed forecasting tools that aid in strategic planning within the renewable energy industry. The usefulness of a functional data horizon in improving prediction reliability is further supported by these findings.