Improving Prediction Accuracy in Drift Detection Using Random Forest in Comparing with Modified Light Gradient Boost Model
N. Mohana Likitha, T. J. Naga Lakshmi · 2024
Improving the comparison of the Random Forest model's predictive power in drift detection with that of the modified light gradient boost model is the ultimate goal of this suggested study endeavor. Research Tools and Procedures: Based on the size of the data set, 40 samples were gathered using different testing and training methods. A combination of Random Forest and a modified light gradient boost model was used to assess the rising prediction accuracy in drift detection. The accuracy was tested using these models. There are a total of 40 samples, split evenly between two groups: Random Forest the Light Gradient Boost Model with Some Adjustments. It is assumed that the G power is 80 %. Findings: The simulation's mean Random Forest accuracy was 0.71 and the Modified Light Gradient Boost Model accuracy was 0.96708. A 2 tailed significance level of 0.0 was derived, which is lower than the threshold of$\mathbf{p} < 0.05$. The results demonstrate a statistically significant relationship between the two categories. Conclusion: Compared to Random Forest outperforms the Modified Light Gradient Boost Model with much lesser accuracy when tested on the given dataset.