Simulated Event Information Generation Using Grid Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating simulated occurrence patterns of events, such as kernel density estimation, are time-consuming and may not accurately reflect actual event occurrences due to uniform interpolation methods.
Innovation Solution
An information generation device and system that acquire past event information, including occurrence positions and supplementary data, to generate simulated event information by simulating event occurrences in a predetermined area, with occurrence positions determined based on the acquired information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If kernel density estimation is performed to estimate two-dimensional distribution of occurrence probabilities, then the distribution can be estimated, but it takes time to calculate
Solution Approach 1:
The patent segments the continuous space into discrete grid cells and divides the calculation into multiple processing stages: (1) reading past event data and creating initial grid data, (2) performing kernel density estimation on the grid data to calculate occurrence probabilities, and (3) generating simulated occurrence patterns from the probability distribution. This segmentation allows for more efficient processing compared to continuous space calculations.
Solution Approach 2:
The patent performs preliminary processing by pre-calculating the two-dimensional distribution of occurrence probabilities from past event data before generating simulated occurrence patterns. The grid data and probability distributions are prepared in advance, which enables faster generation of multiple simulated patterns without repeating the heavy calculation work each time.
2Ease of manufacture
If a single interpolation method such as Gaussian distribution is applied uniformly, then the process is simple, but actual event occurrences are not appropriately reflected
Solution Approach 1:
The patent applies local quality by allowing different interpolation methods to be used in different regions or for different types of events. Instead of forcing a single Gaussian distribution uniformly across all data, the system can select appropriate interpolation methods based on local characteristics of the event data, such as different kernel functions for different event types or regions with different spatial patterns.
Solution Approach 2:
The patent enables parameter changes by allowing the interpolation method parameters to be adjusted based on the characteristics of the past event data. The system can modify kernel bandwidth, kernel function type, and other parameters to better fit the actual distribution patterns observed in the data, rather than using fixed uniform parameters.
Data Source
AI summary
An information generation device acquires past event information including information on a position where a predetermined event occurred and supplementary information regarding the position, and generates simulated event information obtained by simulating occurrence of the event in a predetermined area, with an occurrence position of the simulated event information being a position in the predetermined area based on the position included in the past event information and the supplementary information regarding the position.