Energy resources on the Earth are limited which makes renewable energy a crucial requirement to ensure the utilization of non-renewable resources at a sustainable rate. To address this problem, previously, a low cost, energy efficient hardware model was presented which was capable of floating in air and regularly sending the information related to monitored weather conditions to cellular phone using gsm networks [2]. By virtue of the recorded values they predicted manually taking an approximation that whether the place is favorable for establishment of solar and wind energy generation plants. In this paper, we intend to apply machine learning techniques that can operate on a much larger dataset to identify the probable sites for establishing solar energy power plant efficiently. We compare multiple classification techniques like nearest centroid classification, stochastic gradient descent (SGD) classification and decision tree classification for generating the most effective model. The dataset for experimentation was gathered from the model that was proposed in [2]. The results indicate that decision tree approach resulted in the highest accuracy of over 99.8 % and outperformed nearest centroid and stochastic gradient descent (SGD) approaches for the mentioned dataset.