基于GS-LGBM的乘用车使用寿命预测研究

为准确预测乘用车使用寿命，提出基于网格搜索优化LightGBM（GS-LGBM）模型的乘用车使用寿命预测方法。通过对2014—2019年乘用车报废数据进行大量实验，并与9种流行的机器学习算法进行对比，结果表明，LightGBM在平均绝对误差（MAE）、中位绝对误差（MEAE）、均方误差（MSE）和拟合优度判定系数（R2）4项指标上均明显优于其他算法。为进一步提升模型预测精度，采用网格搜索算法对LightGBM进行参数优化构建GS-LGBM模型,效果显著提升（MAE降低11.02%），说明该方法能够更准确高效地预测乘用车使用寿命。

Research on Life Prediction for Passenger Vehicles Based on GS-LGBM

A life prediction method for passenger vehicles based on the LightGBM model optimized by grid search (GS-LGBM) is proposed to accurately predict the useful life of passenger vehicles. A large number of experiments were conducted on passenger vehicle scrap data from 2014 to 2019, and the comparisons were made with four single traditional machine learning algorithms and five ensemble algorithms based on decision trees. The results show that LightGBM is superior to other algorithms in Mean Absolute Error (MAE), Median Absolute Error (MEAE), Mean Square Error (MSE) and R-squared (R2). To further improve the prediction accuracy of the model, the GS-LGBM model is established by using grid search algorithm to optimize the parameters of LightGBM and the performance was significantly improved (MAE decreased by 11.02%). The results indicate that this method can predict the useful life of passenger vehicles more accurately and efficiently.