Detection of Dried Cannabis Aroma Using the Random Forest Algorithm

Abstract

Cannabis (Cannabis sativa) is the most widely consumed type of illicit drug, accounting for 56.7% of usage. Its abuse leads to numerous negative consequences, including impaired brain function, which can trigger hallucinations, illusions, cognitive disturbances, sudden mood swings, and addiction. The high prevalence of such cases presents challenges, including the high cost of equipment used for detection. To address this, an "E-Nose" (electronic nose) system has been designed to mimic the human sense of smell. This research involves processing collected cannabis data to develop a detection model by comparing two classification methods—the Random Forest and Decision Tree algorithms to determine which yields the highest classification accuracy. The Random Forest algorithm achieved an accuracy of 100%, successfully distinguishing between samples of dried cannabis and non-cannabis substances. The Decision Tree algorithm, meanwhile, achieved an accuracy of 70% in correctly identifying dried cannabis.