An Enhanced XGBoost Algorithm for Mobile Price Classification

With the popularity of mobile devices in recent years, understanding and predicting mobile prices is crucial for mobile manufacturers. Mobile price can be influenced by different hardware factors, making accurate prediction of prices complex and challenging. In this paper, a novel classification model combining Dung beetle optimizer (DBO) and XGBoost is proposed to accurately classify mobile price. In the PCA dimensionality reduction phase, we design a feature filtering strategy with consideration of feature importance. Through DBO, the optimal parameters of XGBoost are selected to build the model. The experimental results demonstrate the superiority of our proposed DBO-XGBoost model compared with the baseline models such as standard XGBoost, Decision Tree (DT), Random Forest (RF), and AdaBoost models. Our model scores 95.5% in classification accuracy with the kaggle mobile-price-classification dataset, outperforming the other models (XGBoost: 91.8%, DT: 85.3%, RF: 90.5%, AdaBoost: 71.5%). Furthermore, the performance of DBO is verified to be better than grid search for parameter optimization.