<![CDATA[The mixing of pure coconut oil (VCO) with cheaper vegetable oils has negative impacts on both consumers and producers. This study aims to develop a method for detecting VCO adulteration using an ESP32-based dielectric sensor combined with a Random Forest classification algorithm. The research employed an experimental design using 225 samples, including pure VCO, canola oil, corn oil, and various mixture ratios, each measured with five repetitions. The results show that pure VCO exhibits the highest capacitance values (58.4–62.4 pF), followed by canola oil (44.8–47.8 pF) and corn oil (43.2–46.6 pF), indicating clear differences in dielectric properties related to fatty acid composition. ANOVA analysis confirmed a significant difference between pure VCO and adulterated oils (p < 0.05). The Random Forest model achieved an accuracy of 53–58% for 15-class classification, while binary classification (pure vs adulterated oil) reached more than 90% accuracy. This finding is discussed in terms of the effectiveness of dielectric sensing combined with machine learning for distinguishing oil authenticity. In conclusion, the proposed system provides a fast, low-cost, mobile, and user-friendly solution for VCO quality monitoring, with potential applications in supply chain control and consumer protection.]]>
Copyrights © 2026
