Commuting time prediction method, training method and device for commuting time prediction model

Abstract

This application proposes a commuting time prediction method, a commuting time prediction model training method and device, wherein the method includes: receiving the departure place, destination and departure time; mapping the departure place and destination to the road network to obtain road network sections Sequence number and section offset; map the departure time to the time slice to obtain the time slice number and time slice offset; according to the time-space input encoding module, the road network section number, section offset, time slice sequence number and time slice offset The fixed-length vector is obtained by encoding the quantity; the fixed-length vector is processed according to the time prediction module, and the target commuting time is output. As a result, the accuracy of commuting time prediction from the departure place to the destination is improved.

Description

Technical field [0001] The present application relates to the field of information retrieval, and in particular, there is a training method and apparatus for a commuter time prediction method, a commuter time prediction model. Background technique [0002] In the field of transportation, commuter time predictions are an important topic that plays an important role in many transportation services, such as map navigation recommendation services and urban logistics services. Commuter Time prediction refers to a given departure, destination, and departure time, and then predicting that the starting time begins how long it takes from the departure point to the destination. Further, the prediction problem can be divided into two major categories: one is a given path prediction time, and the other is not given a specific path prediction time. [0003] In the relevant technology, the problem is also called OD (Origin-Destination, departure - destination) Commuter time prediction problem,...

AI Technical Summary

Problems solved by technology

Therefore, it becomes a challenge to solve the OD commuting time prediction problem if the OD input is bound to the selected path without actually traversing to generate all the paths.

Images