Evaluation of Vehicle Quality Performance using Random forest in Comparison with KNN to measure the Accuracy, Recall, and Precision
V. Ramya, Kirupa Ganapathy · 2022 3rd International Conference on Intelligent Engineering and Management (ICIEM) · 2022
The major goal of this paper is to detect the performance of Random Forest algorithm in detection of vehicle quality performance by equating it with the K-Nearest Neighbor algorithm (KNN). Materials and Methods: Proposed work uses a total sample size of 1208, and the dataset is collected from UCI repository with accurate quality 604 samples and inaccurate quality 604 samples with two groups. Group 1 with RF algorithm and group 2 with KNN. The samples are separated into training data (n = 906 [75%]) and test data (n = 302 [25%]). Calculation of samples is done using G power analysis with alpha (0.05), power (80%), and environment ratio are two different groups in clincalc. Results: In the proposed model Random Forest achieved accuracy, recall and precision of 90.38%, 92.53% and 82.08% respectively compared to 83.33%, 91.83% and 79.25% by the KNN algorithm with a significance of 0.000 (p<0.05). Conclusion: In this research, Random Forest shows better results than the KNN algorithm in car evaluation dataset with more features considered.