Geolocation Data Normalization for Representative Visitor Counts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Geolocation data often lacks accuracy and reliability due to inconsistent recording methods, third-party processing, privacy laws, and undercounting of user devices with low observation frequency, leading to inaccurate representation of user behavior.
Innovation Solution
A system that normalizes geolocation data by using a computer network to analyze user device observations, adjusting for dwell time, observation frequency, and demographic factors, and applying probability functions to correct undercounting, thereby determining a more accurate number of visitors at specific locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If location data is recorded through various applications and devices, then data coverage and accessibility are improved, but data accuracy and reliability deteriorate due to inconsistent recording methods and sampling intervals
Solution Approach 1:
The system applies parameter changes by adjusting the observation probability model based on dwell time parameters. Devices are weighted according to their observed dwell times at points of interest, transforming raw count data into normalized estimates that account for varying observation frequencies and durations, thereby resolving the contradiction between data coverage and measurement precision
2Productivity
If third-party services process location data in various ways, then data processing capability is improved, but data representativity deteriorates due to processing biases and inconsistencies
Solution Approach 1:
The system introduces an intermediary normalization layer that sits between raw location data and final analytics. This intermediary process applies probability models and weighting factors to adjust data representativity, compensating for biases introduced by third-party processing while maintaining the productivity benefits of centralized data handling
3Object-affected harmful factors
If privacy laws affect location data recording, then data privacy protection is improved, but data completeness deteriorates due to selective data collection
Solution Approach 1:
The system addresses privacy-induced data completeness issues by changing the analysis parameters to account for selective observation. The probability model adjusts for devices that may be under-observed due to privacy settings, allowing accurate estimates despite incomplete data collection caused by privacy protections
4Adaptability or versatility
If devices with low observation frequency are included, then device diversity is improved, but data reliability deteriorates due to insufficient observation samples
Solution Approach 1:
The system applies parameter changes through the probability model that adjusts weighting based on observed dwell times. Devices with low observation frequency are included in the analysis but are assigned appropriate weights that reflect their limited data contribution, maintaining device diversity while preserving data reliability through statistically sound weighting
Data Source
AI summary
A system and method for normalizing device counts from recorded device observation data has been developed. The observation data includes locations of the devices recorded at various points in time. The system normalizes the device counts to account for devices that are not accurately represented in the data. The method includes calculating a probability that a device is observed at a location based on the observation frequency of the device and the dwell time at that location. The method further includes calculating a normalization factor based on the population of the geographic region that the device is located. In one example, the method further includes calculating a number of visitor devices and/or a number of overnight visitors at a location.