Footfall Estimation via Visit Data Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing footfall estimation systems face challenges in accurately estimating the number of people present in a location due to inconsistencies in user visitation data, such as varying usage patterns of mobile apps and differences in GPS signal quality across indoor and outdoor locations.
Innovation Solution
The system clusters visitation data into groups of visits based on time and location conditions, and then applies a machine learning model to these clusters to determine a more accurate footfall estimation, reducing the impact of inconsistent data representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS signals are processed to obtain user visitation data, then geographical precision is improved, but data inconsistency across different locations and times worsens
Solution Approach 1:
The patent segments the mass GPS data into individual user visitation records, each with unique identifiers. This segmentation allows for tracking and analyzing each user's movement pattern separately, enabling the system to handle the inconsistency by processing each record individually through the machine learning model rather than treating all data uniformly.
Solution Approach 2:
The patent applies a machine learning model that dynamically adjusts parameters based on user behavior patterns, location characteristics, and time variations. This allows the system to adapt to the inherent inconsistency in GPS data by changing processing parameters according to specific conditions, thereby maintaining reliability despite varying data quality across different locations and times.
2Ease of operation
If all GPS records are processed equally, then processing simplicity is maintained, but footfall estimation accuracy deteriorates
Solution Approach 1:
The patent performs preliminary actions by training a machine learning model in advance using historical GPS data. This pre-trained model captures user behavior patterns and location characteristics, enabling accurate footfall estimation without complex real-time processing. The simplicity is maintained during operation while accuracy is improved through the preliminary training phase.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between raw GPS records and footfall estimation. This intermediary processes the GPS data according to learned patterns rather than treating all records equally, thereby improving accuracy while keeping the overall system operation simple through automated model-based processing.
3Quantity of substance
If mass GPS data is stored for analysis, then data availability is improved, but storage requirements increase
Solution Approach 1:
The patent extracts only the essential elements from mass GPS data - user identifiers, location coordinates, and timestamps - and stores these refined visitation records. By taking out only the necessary data elements needed for footfall estimation rather than storing complete raw GPS datasets, the system maintains data availability while significantly reducing storage requirements.
Data Source
AI summary
The presently disclosed subject matter includes a computerized method and system for estimating footfall for a location of one or more users. Visitation data including a plurality of visit data items is obtained. The obtained visitation data is grouped into one or more groups of visit data items. For each group, the data items in the group is assigned to one or more clusters of visits. For each visit data item, the data item is assigned to a given cluster of the clusters. Based on the one or more clusters of visits data, a footfall estimation for a location is determined, wherein the footfall estimation for the location is indicative of a number of visits of the one or more users in the location, within one or more time intervals.