Count Regression Model for Transportation Incident Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Public transportation systems face challenges in identifying the causes of service unreliability and predicting future incidents, which affects schedule adherence and operational costs, due to the inherent variability in demand, operator performance, traffic, weather, and other unpredictable factors.
Innovation Solution
A method and system using count regression models based on transportation incident data, including a contingency table analysis, to predict future incidents and adjust operational parameters, such as vehicle schedules and resource allocation, by constructing count regression models from CAD/AVL system data to identify patterns and relationships between variables like time, day, and month, and using Poisson regression equations to forecast incident occurrences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring systems are used to track transportation incidents, then data collection is simple, but the ability to predict future incidents and identify causes is insufficient
Solution Approach 1:
The system performs preliminary analysis by constructing contingency tables and regression models using historical incident data before future incidents occur. This advance modeling enables prediction of future incidents by identifying patterns and relationships in past data, allowing transportation providers to take preventive actions before problems arise.
Solution Approach 2:
The patent introduces contingency tables and regression models as intermediary analytical tools between raw incident data and prediction outcomes. These intermediaries process and transform raw data into meaningful patterns and predictions, bridging the gap between simple data collection and accurate incident forecasting.
2Measurement precision
If detailed incident data is collected for every event, then prediction accuracy improves, but data processing complexity and time increase
Solution Approach 1:
The system segments the analysis process into distinct stages: data collection, contingency table construction, regression model development, and prediction generation. This segmentation allows each stage to be optimized independently and enables parallel processing of multiple incident types simultaneously, reducing overall processing time while maintaining analytical depth.
Solution Approach 2:
The patent applies partial action by focusing regression analysis on specific incident types and time periods rather than analyzing all data uniformly. This selective approach concentrates computational resources on the most critical incidents, achieving high prediction accuracy for key problems without the excessive time cost of comprehensive analysis of every data point.
3Adaptability or versatility
If regression models are constructed for all incident types, then comprehensive prediction coverage is achieved, but computational resources and complexity increase
Solution Approach 1:
The patent develops a universal regression modeling framework that can be applied to multiple incident types using the same underlying methodology. This universal approach provides comprehensive prediction coverage across different incident categories while avoiding the complexity of developing separate specialized models for each incident type, as the same contingency table and regression techniques serve multiple purposes.
Data Source
AI summary
A method and a device for predicting a future occurrence of a transportation system incident are disclosed. The method includes receiving transportation incident data comprising information collected during and related to operation of at least one transportation vehicle, the information comprising at least a time stamp, constructing at least one count regression model based upon the transportation incident data, producing a results set based upon the at least one count regression model, predicting at least one future occurrence of an incident based upon the results set, and presenting the at least one predicted future occurrence. The device includes at least a processing device and computer readable medium containing a set of instructions configured to cause the device to perform the method.