Toward Prediction of Traffic Accidents Using Formal Concept Analysis of Actual Accidents and Related Data

Abstract

This study uses Formal Concept Analysis (FCA) to investigate factors of traffic accidents by analyzing actual traffic accident data including its date, place, injury severity, road shape, accident summary in a natural language, etc for each accident. FCA is a mathematical theory of data analysis based on formal contexts and concept lattices. We gather data related to each of the traffic accidents such as land use districts, traffic volumes, and so on, translate them into a binary context table as an input of FCA, and analyze conceptual structures as an output of FCA to investigate traffic accident factors.