摘要
提出一种基于LightGBM的多阶段医疗服务等待时间的预测方法。结合文献调研找出患者等待时间的影响因素,根据影响因素采用SIMIO仿真软件收集相关训练数据,通过独热编码和数据标准化进行数据预处理,然后采用Lasso、Ridge、GBDT、XGBoost和LightGBM(Light Gradient Boosting Machine)构建预测模型,并用随机搜索进行参数寻优。综合寻优时长和预测精度,LightGBM算法消耗较短的寻优时长同时获得较高预测精度,平均绝对误差为3.439 1,平均百分比误差为8.52%。
This paper proposes a prediction method of waiting time for multi-stage medical services based on LightGBM. We combined with literature research to find out the influencing factors of patients’ waiting time. According to the influencing factors, SIMIO simulation software was used to collect relevant training data. We performed the data processing through one-hot encoding and data standardization. Lasso, Ridge, GBDT, XGBoost and LightGBM was used to construct the prediction model, and the random search was applied to optimize the parameters. Considering the long optimization time and prediction accuracy, LightGBM algorithm consumes less optimization time and obtains higher prediction accuracy. The average absolute error is 3.439 1, and the average percentage error is 8.52%.
作者
彭俊
项薇
谢勇
黄益槐
韩乐奇
吴成宇
Peng Jun;Xiang Wei;Xie Yong;Huang Yihuai;Han Leqi;Wu Chengyu(Faculty of Mechanical Engineering and Mechanics,Ningbo University,Ningbo 315211,Zhejiang,China;Institute of Advanced Energy Storage Technology and Equipment,Ningbo University,Ningbo 315211,Zhejiang,China;Fuzhou Preschool Education College,Fuzhou 344099,Jiangxi,China)
出处
《计算机应用与软件》
北大核心
2022年第12期119-124,共6页
Computer Applications and Software
