Classifying food according to their cooking degree with electronic nose

One of the most commonly used cooking methods in daily life is baking in the oven. This method is highly preferred because food loses less nutritional value and is healthier and more practical than other methods. In this study, research was conducted to make it easier to follow the cooking levels of oven-baked foods and to prevent overcooking of foods. In this thesis study; 4 types of food, including cakes, pastry, horse mackerel and salmon, cooked in the built-in oven, are classified into 3 classes: undercooked, cooked and overcooked, according to their cooking degrees, with the electronic nose circuit designed using 8 MOS Metal Oxide Semiconductor sensors. The smell of the food cooked in the oven is detected with the help of sensors and converted into electrical signals. These electrical signals were recorded with the help of a data acquisition card and digitized in a computer environment. First, pre-processing was applied to the obtained signals and feature extraction was made. Then, 4 different normalization methods were applied to the feature data: z-score, maximum, minimum-maximum and sigmoid. In order to test the effect of different normalization methods on classification, data without and with normalization were classified using decision tree, k-nearest neighbors, support vector machines and random forest algorithms. Classification accuracies were examined with the 5-fold cross-validation method and success percentages were compared.