Optimal Feature Selection Algorithm (FSA) for Electronic Nose Signal

In E-Nose data, redundancy and overlapping are frequent issues. This is due to the fact that certain sensors have the same gas target as other sensor gas targets. The goal of this research is to classify meat odor E-Nose data into four classes, including chicken, beef, lamb, and pork. 276 meat odor data in total–108 chicken, 61 beef, 27 lamb, and 80 pork–were used in this research. Additionally, this research aims to identify the most effective Feature Selection Algorithm (FSA) to address issues with redundancy and overlapping data. In this study, FSA techniques such as Pearson Correlation, ANOVA F -value, Chi-Square, and MIFS were applied. SVM, K-NN, and Random Forest are the implementations of classification algorithms. According to the research's findings, SVM-MIFS (k = 10) is the best combination method for classifying meat odor data, achieving an accuracy rate of 98.91 %.