A Multiple Linear Regression Model to Estimate Global, Direct and Diffuse Irradiance in Gurugram, India, Using Python

Estimating solar radiation is important for designing a solar photovoltaic or thermal system. In the present work, multiple linear regression model is used to estimate global horizontal irradiance, direct normal irradiance and diffuse horizontal irradiance using the solar resource assessment data collected from National Institute of Solar Energy located in Gurugram (28.42° N, 77.15° E) which records direct, diffuse, global radiation at an interval of 1 min along with humidity, wind speed, wind direction, temperature and precipitation. Principal component analysis is used to select the dominating variables and then fit a multiple regression model to estimate the components of solar radiation. The model performance is tested by computing the coefficient of determination, root mean square error, mean bias error, mean absolute error and model efficiency for each of the models. Model efficiency of 0. 93, 0.88 and 0.93 respectively are obtained for the multiple regression models for estimating global, direct and diffuse irradiance which suggests that the model fits well with the observed data.