模型训练方法、预测方法、装置和存储介质
By selecting historically predicted driving routes with high similarity and training the model by combining real-time and historical road features, the problem of inaccurate prediction of low-speed driving time for ride-hailing services has been solved, resulting in more accurate pricing and improved user experience.
CN115239359BActive Publication Date: 2026-07-17ALIBABA INNOVATION PRIVATE LIMITED
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ALIBABA INNOVATION PRIVATE LIMITED
- Filing Date
- 2021-04-25
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing ride-hailing service platforms often fail to accurately predict low-speed travel time when estimating prices, leading to inaccurate pricing and impacting user experience.
Method used
By acquiring historical estimates and actual driving routes from historical trips, target historical estimated driving routes that meet the similarity threshold are selected. Combining real-time and historical road condition features, a machine learning model is trained to predict low-speed driving duration.
Benefits of technology
It improves the accuracy of low-speed driving time prediction, ensures the accuracy of estimated prices, and enhances the user experience.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN115239359B_ABST
Abstract
本发明实施例提供了一种模型训练方法、预测方法、装置和存储介质。所述模型训练方法包括：获取原始样本，原始样本包括历史行程的历史预估行驶路线及相应的历史实际行驶路线；确定路线相似度满足预设的相似度阈值的历史预估行驶路线作为目标历史预估行驶路线；获取目标历史预估行驶路线包含的目标历史道路的实时道路状态特征作为实时类特征；获取目标历史道路的历史道路通行统计特征作为历史类特征；采用实时类特征和历史类特征作为训练样本，对预设的机器学习模型进行训练，以得到能用于预测低速行驶时长的预测模型。在本发明实施例的方案中，训练得到的预测模型能够基于预估行驶路线准确地预测低速行驶时长。
Need to check novelty before this filing date? Find Prior Art