Energy is inextricably tied to the health condition. Various countries do not require the labelling of nutrition information on packaged food products. This lack of information leaves consumers and policymakers unaware of the energy content (low energy content, high energy content) in these products. To address this issue, we have created a machine learning-based approach that employs nutrition facts labels to estimate the energy content of packaged food products. We obtained 204 samples of nutrition information, of which (n=152) samples were utilized for the training dataset (and n=52) samples were utilized for testing. This approach enables complete traceability also enhance the clarity and precision of food product labeling. Utilizing of various machine learning algorithm (SVM, KNN, RF, DT, MLR) and measure the performance KNN performed the highest accuracy of 97.06%. our research emphasizes the potential of applying machine learning to effectively forecast the energy value of packaged food in a large scale.