부산대학교 도서관

Using two different classifiers—an adaptive boosting (AdaBoost) and a novel extreme gradient boosting (XGBoost)—this study compares and contrasts the two in order to draw conclusions about online personas. Materials and Methods: A publicly available dataset from Kaggle is the main source of data used in this study. The study used a sample size of twenty people, split evenly between two groups of ten, in an effort to detect personality traits. Using a 95% confidence interval, an alpha of 0.05, and a beta of 0.2, the G-power 0.8 was employed to ascertain the suitable sample size. In this work, the researchers examine the prediction of personality traits using two distinct machine learning algorithms: XGBoost and AdaBoost. There are a total of 20 participants in both groups, with 10 participants assessed for each algorithm. Finding out if these ML systems can successfully forecast personality traits from the provided dataset is the primary goal of the research. The study's findings might influence how researchers in the area of personality prediction and machine learning algorithms go about their work in the future. Results: The results demonstrate that compared to the AdaBoost classifier's 93.20% accuracy rate, the XGBOOST algorithm has a far higher prediction accuracy of 98.55 percent. This study's p value, which indicates statistical significance, is 0.039. The study's findings and the p=0.001 indicate that the two algorithms are significantly different. The hypothesis is proven valid and an independent sample T-test was used, as indicated by the significance value of 0.001 (p<0.005). Conclusion: When comparing the two classifiers, XGBoost is more accurate in predicting people's personalities than AdaBoost, which stands for adaptive boosting. [ABSTRACT FROM AUTHOR]