Classification-Based Authentication between Fresh Cooking Oil and Bleached Waste Cooking Oil Using an Electronic Nose

Consuming too much leftover cooking oil can cause diabetes, obesity and cancer, among other health problems. Used cooking oil usually has different physical and odor characteristics than fresh cooking oil. However, when purified with bleaching earth and activated carbon, used cooking oil can take on physical characteristics identical to those of fresh cooking oil, making visual and olfactory differentiation difficult. This provides an opportunity for illegal businesses to resell used cooking oil as new, posing a health risk to consumers. Experiments have shown that machine learning algorithms can assist in classifying oil into two categories as bleached waste cooking oil and fresh cooking oil. Among the methods tested, the Random Forest algorithm demonstrated the best performance, achieving an accuracy of 0.9947, precision of 0.9950, recall of 0.9947, and an F1 score of 0.9948. These results indicate that the random forest algorithm is effective in supporting an electronic nose system in identifying and distinguishing bleached waste cooking oil from fresh oil. Thus, this system offers an effective solution for monitoring cooking oil quality, helping to prevent the distribution of potentially harmful waste cooking oil in the market.