Crowd Movement Mining via Trajectory Segmentation and Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Organizations face challenges in understanding crowd movements and making informed decisions about urban planning and resource allocation due to the lack of effective methods for collecting and analyzing location data from user devices.
Innovation Solution
A method that collects location data from user devices, mines frequent patterns to generate representative sequences, and groups these sequences into clusters based on sequence distances to identify crowd movements, using techniques such as sequential pattern mining and agglomerative hierarchical clustering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If location data is collected from multiple user devices, then crowd movement information can be obtained, but data processing complexity and computational resources required increase
Solution Approach 1:
The patent segments the crowd movement analysis by dividing individual trajectories into discrete segments (start point, end point, duration) and organizing them into patterns. This segmentation allows complex movement data to be processed in manageable units, reducing overall processing complexity while preserving comprehensive movement information.
Solution Approach 2:
The patent extracts key features from raw location data, such as start points, end points, and time durations, to create simplified trajectory representations. By taking out only the essential information needed for crowd movement analysis, the system reduces data complexity while maintaining informative content.
2Measurement precision
If frequent patterns are mined to generate representative sequences, then crowd movement patterns are identified, but computational time and processing resources increase
Solution Approach 1:
The patent applies partial action by focusing pattern mining on the most significant and frequent movement patterns rather than analyzing all possible patterns exhaustively. This approach identifies the most relevant crowd movement patterns efficiently, reducing computational time while maintaining high identification accuracy for meaningful movements.
Solution Approach 2:
The patent changes parameters such as minimum support thresholds and pattern length criteria to optimize the balance between computational time and pattern identification accuracy. By adjusting these parameters, the system can quickly identify representative sequences without excessive computational resources.
3Adaptability or versatility
If representative sequences are grouped into clusters, then crowd movements are categorized, but algorithm complexity and processing steps increase
Solution Approach 1:
The patent merges similar representative sequences into clusters based on their characteristics, creating categorized crowd movement groups. This merging process simplifies the data structure by organizing individual trajectories into meaningful categories, reducing overall complexity while enhancing categorization capability.
Solution Approach 2:
The patent uses clustering algorithms that can be configured with different parameters (such as distance thresholds and cluster numbers) to adapt to different crowd movement scenarios. By adjusting these parameters, the system achieves versatile categorization without requiring overly complex algorithms for each specific case.
Data Source
AI summary
A method for finding crowd movements is provided. The method includes: collecting location data associated with multiple user devices, mining frequent patterns in the location data to generate multiple representative sequences, and grouping the representative sequences into clusters to find crowd movements according to sequence distances between the representative sequences. Each representative sequence includes at least one line segment between a starting location point and an end location point.