Personalized Route Prediction via Evolving Database Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions for predicting personalized and contextually relevant mobility patterns require iterative processing of historical data, leading to inefficiencies due to the need for re-training and are unable to produce both simple and complex, high-precision prediction outputs, especially as user needs and habits evolve over time.
Innovation Solution
A system that uses evolving databases to decompose trip data by date and time partitions, creating Markov tables and full route databases to learn and predict departure points, routes, and destinations with one-pass processing, allowing for context-aware adjustments and geospatial and temporal pattern encapsulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative processing with re-training is used, then prediction accuracy can be improved, but processing time and computational efficiency deteriorate
Solution Approach 1:
The patent segments the trip data processing into distinct components: extracting trip data, decomposing by date/time partitions, creating Markov tables, and generating route predictions. This segmentation allows each component to be processed independently and efficiently, eliminating the need for iterative re-training while maintaining prediction accuracy through structured data decomposition and specialized database creation.
2Loss of information
If raw data logging is used, then data completeness is improved, but storage requirements and processing complexity deteriorate
Solution Approach 1:
The patent extracts only the necessary trip data elements (starting point, ending point, route segments, date, time) from raw mobility data and stores them in specialized databases organized by date and time partitions. This extraction approach maintains data completeness for prediction purposes while significantly reducing storage requirements and processing complexity by filtering out redundant information.
Solution Approach 2:
The patent performs preliminary decomposition of trip data by date and time partitions before creating Markov tables and generating predictions. This preliminary action organizes data in advance, reducing the complexity of subsequent processing and enabling efficient one-pass prediction without requiring complex iterative algorithms.
3Ease of operation
If simple prediction outputs are generated, then ease of use is improved, but prediction precision deteriorates
Solution Approach 1:
The patent creates a dynamic prediction system that can adapt its output complexity based on needs. The Markov tables and route databases can provide simple high-level predictions (e.g., destination categories) or detailed precise predictions (e.g., specific route segments, timing) depending on the query requirements, allowing the system to adjust between ease of use and prediction precision dynamically.
Data Source
AI summary
Systems and methods for personalized route prediction including receiving, at a first time, first input data associated with a first route; populating a first database with the input data; receiving, at a third time, second input data associated with a second route; comparing the second input data to the first input data included within the first database; determining, based on the comparison, a first cluster including the first data and the second input data or a second cluster including the second input data; populating a second database based on the first cluster or the second cluster; determining, using the first database and at a second time, at least one of: predicted departure data, predicted destination data, and/or predicted route data; and causing, based on the predicted departure data, predicted destination data, and/or predicted route data, to perform an action in association with a vehicle.