Comparison of Supervised Learning Algorithms for Cigarette and Vape Smoke Classification Using Electronic Nose

This research discusses applying the Supervised Learning method using Electronic Nose to classify the types of cigarette and vape smoke in the air. Electronic Nose is used as a scent detector that can identify the characteristics of smoke from both sources. Three Supervised Learning algorithms, namely KNN, SVM, and Decision Tree, were applied to compare the performance in classifying smoke types. The data comprised reference air samples, air contaminated by manufactured cigarette smoke, rolled cigarettes, and vape. The results showed that all three Supervised Learning algorithms successfully provided an excellent classification for cigarette and vape smoke types using data from Electronic Nose. The best accuracy result was achieved by SVM, with an accuracy rate of 96.55%. This research contributes to identifying sources of air pollution that have the potential to endanger human health