Destination Prediction Using Movement Trends
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing destination prediction techniques fail to accurately predict user destinations when no information on past movement trajectories is available, particularly in situations like first-time visits to new locations or when users have not used services before.
Innovation Solution
A destination prediction device and method that estimates remaining user movement based on current movement trajectories and movement trends among multiple users, using a personal remaining movement estimation unit and a personal destination estimation unit, which combines movement trend information to predict destinations even without past trajectory data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional destination prediction techniques using past movement trajectories are applied, then prediction accuracy is improved for users with historical data, but prediction capability deteriorates for users without past trajectory information
Solution Approach 1:
The prediction system is segmented into two independent components: a personal prediction module that uses individual past trajectories, and a general prediction module that uses aggregated movement trends from multiple users. This segmentation allows each module to operate independently, so new users can immediately use the general module without waiting for personal data accumulation.
Solution Approach 2:
Aggregated movement trends serve as an intermediary between individual users and the prediction system. Instead of requiring direct personal history, the system uses these trends as a mediator to provide predictions for new users, bridging the gap between lack of personal data and prediction needs.
2Adaptability or versatility
If movement trajectories of multiple users are collected and used, then prediction capability for new users is improved, but privacy protection deteriorates
Solution Approach 1:
The system extracts only the necessary aggregated movement patterns from individual trajectories, separating the useful predictive information from personal identifying details. By taking out only the essential movement trends and discarding personal identifiers, the system maintains prediction capability while reducing privacy risks.
Solution Approach 2:
Different levels of data processing are applied: individual user data is processed to extract general trends, while personal identifiers are removed. The aggregated trends are then used for all users including new users, applying a uniform privacy protection standard while maintaining local prediction accuracy where personal data is available.
Data Source
AI summary
A personal remaining movement estimation unit (12) estimates information relating to a remaining movement of a user on the move based on a movement trajectory of the user on the move, a personal destination estimation unit (16) predicts a destination of the user on the move based on information estimated by the personal remaining movement estimation unit (12) and information indicating the number of people moving between areas of a plurality of users stored in a movement trend information storage unit (34), and it is thereby possible to predict a destination even for a user having no information on past movement trajectories.