Comparative Predictive Analysis through Machine Learning in Solar Cooking Technology


  • Karankumar Chaudhari Department of Mechanical Engineering, G H Raisoni College of Engineering, Nagpur, India
  • Pramod Walke Department of Mechanical Engineering, G H Raisoni College of Engineering, Nagpur, India
  • Sagar Shelare Department of Mechanical Engineering, Priyadarshini College of Engineering, Nagpur, India. Centre for Research Impact and Outcomes, Chitkara University, Rajpura, Punjab, India



Solar Cooking, Machine Learning, regression analysis, XGBoost, Statistical Analysis


Renewable energy technology has helped solve global environmental issues in recent years. Solar cooking technology is a sustainable alternative to conventional cooking, particularly in regions with ample sunlight. Although there is a growing interest into solar cooking, however, there is a lack of comprehensive comparison research upon the machine learning models predictive accuracy. Prior studies frequently concentrate upon individual models or fail to conduct comprehensive comparative analyses, resulting in a knowledge deficit regarding the most effective predictive methodologies for solar cooking technology. This research article compares solar cooking with special types of cooking utensils used for indoor cooking by predictive analysis of different kinds of machine learning models. To achieve proper cooking, the temperature of both pan and pot is to be monitored constantly. For this, a machine learning (ML) system model was constructed for predicting pan and pot temperature as a response parameter. By leveraging datasets encompassing time duration of the cooking, mass flow rate of heat transfer fluid, type of heat transfer fluid, and global solar radiations, a range of machine learning algorithms, including decision tree regressor, linear regression, extreme gradient boosting, and random forest regressor algorithms, are employed for predicting pan and pot temperature of solar cookers. Extreme gradient boosting is the best machine learning model for solar utensil temperature, with maximum R2 and minimum mean squared error, mean absolute error, and root mean squared error values that perfectly predict all answers. Also, extreme gradient boosting predicts well on training and testing datasets, whereas Random forest predicts well on training datasets but poorly on test data, causing overfitting. This research shows that machine learning could revolutionize solar cooking technology, promising a future for renewable energy and sustainable living.




How to Cite

K. Chaudhari, P. Walke, and S. Shelare, “Comparative Predictive Analysis through Machine Learning in Solar Cooking Technology”, IJECES, vol. 15, no. 6, pp. 543-552, Jun. 2024.



Original Scientific Papers