Improved Accuracy in Detection of Driver Drowsiness Using Novel Random Forest Classifier Compared with Decision Tree Classifier

The goal of this study is to evaluate how well driver drowsiness can be detected using two different machine learning methods: the Decision Tree Classifier and the Novel Random Forest Classifier. Each method was tested using a sample size of size of 10. To determine the effectiveness of each method, we used SPSS to predict the significance value of the dataset, setting the G-power value at 85%. We compared the accuracy of the two classifiers with different sample sizes to see which performed better. The results showed that the Novel Random Forest Classifier outperforms the Decision Tree Classifier in detecting driver drowsiness. Specifically, the Novel Random Forest Classifier achieved an accuracy of 63.20%, while the Decision Tree Classifier reached 51.80.%. This difference is statistically significant, with a significance value of $0.001(p<0.001)$. In conclusion, the Novel Random Forest Classifier is more effective at predicting driver drowsiness compared to the Decision Tree Classifier. It offers a more accurate detection method and can be considered a better option for identifying when drivers are drowsy.