Machine Learning Characterization of Various Snacks in Relation to Their Physico-Chemical and Nutritional Profile Using Electronic Equipment

Using machine learning models, it is easier to determine various important factors relating to the target market and the buyers of snacks and thus develop products to address the needs that are arising from the residents. By using machine learning methods, food manufacturers receive helpful information regarding consumers' preferences and create snacks that are not only tasty but also appropriate for consumption. This study aims to estimate the physico-chemical and nutritional aspects of five different snack types, including deep-fried snacks, extruded snacks, baked snacks, namkeen, and popcorn. These snack foods are evaluated in relation to their moisture content, acid value, peroxide value, fats, protein, carbohydrates, calories, cholesterol, trans, and saturated fatty acids using electronic equipment. This study uses two prediction models which are random forest and bagging algorithms for investigation. The results indicate that baked snacks are the healthiest among the examined options, exhibiting the following nutritional characteristics: The values obtained were: moisture content: 1.68%, ash content: 1.89%, acid insoluble ash: 0.09%, acid value: 0.48, peroxide value: 0.80, protein content: 6.20 g/100 g, fat content: 29.21 g/100 g, Cholesterol content: < 0.1 mg/100 g and fatty snack Baked products are considered to be healthier than other snacks because baked products contain low fats. This study helps to recognize other options of snacks that are more health beneficial in terms of nutritional content. As a result, this study measures the classification performance of two ensemble learning techniques, Random Forest and Bagging, on a number of snack items. The accuracy of these methods was analyzed for five snack categories. The results show Random Forest has higher accuracy compared to Bagging for all snack types.