Temperature effect of electronic nose sampling for classifying mixture of beef and pork

Abstract

Strong demand and strong price of raw foodstuffs like beef was commonly used in conventional markets by beef dealers to commit fraud in order to gain larger income. The fraud has been in the form of combining beef and pork. In Indonesia, this has been a issue of food health in recent years. Via scent, some food safety concerns can be expected. By using electronic nose that is equipped with electrochemical and air sensors such as temperature sensors, strain, and humidity to find the pure beef or mixed beef. According to its selectivity, the sensor can detect gas to make small icurrents that are the result of chemical sensor and gas interactions with oxygen. In this study, the classification method k-NN, SVM, Naïve Bayes, and Random Forest was used in 5 different meat variations with a ratio of 0%, 10%, 50%, 90% and 100% with temperatures of -22° C, Room Temp., And 55° C. The results showed the effect of temperature on increasing the accuracy, which is at a temperature of -22° C. The lower the temperature, the more stable the value obtained by electronic nose. At a temperature of -22° C, the method that produces the highest accuracy is the Random Forest method.