This study comes out with an improved machine learning method for the prediction of wind speed. A new approach for the estimation of missing data using cubic spline function was formulated. Improved models for computing wind power were used taking into account the study area is located in a hot and humid region. The performances of small-scale wind turbines were demonstrated using manufacturers curves and formulated models.