Effective storage and capture of global solar radiation (GSR) is significant as an alternate energy source. The global radiation levels are influenced by various meteorological and air pollutants parameters. Four significant parameters (temperature, relative humidity, wind direction and ozone) were considered to predict global solar radiation using XG-Boost, Random Forest (RF), kNearest Neighbour (k-NN), Support Vector Machine (SVM) and Decision tree (DT) for Alandur provinces of Chennai district, Tamil Nadu. The performance of models was improved through tenfold cross validation. The predicted solar radiation for different ML algorithms yielded an accuracy of 77.42% (XG-Boost), 70.97% (k-NN), 79.26% (RF), 69.59% (DT) and 72.81% (SVM) respectively. Out of the five different MLs, RF algorithm showed best performance and achieved a higher accuracy of 79.26% with precision of 78%, recall and F1 score as 79% respectively for the given air pollutants and meteorological dataset.