DATA FITS-A HETROGENOUS DATA FUSION FRAMEWORK FOR TRAFFIC AND INCIDENT PREDICTION

Abstract

This paper introduces DataFITS (Data Fusion on Intelligent Transportation System), 
an open-source framework that collects and fuses traffic-related data from various 
sources, creating a comprehensive dataset. We hypothesize that a heterogeneous data 
fusion framework can enhance information coverage and quality for traffic models, 
increasing the efficiency and reliability of Intelligent Transportation System (ITS) 
applications. Our hypothesis was verified through two applications that utilized traffic 
estimation and incident classification models. DataFITS collected four data types 
from seven sources over nine months and fused them in a spatiotemporal domain. 
Traffic estimation models used descriptive statistics and polynomial regression, while 
incident classification employed the k-nearest neighbors (k-NN) algorithm with 
Dynamic Time Warping (DTW) and Wasserstein metric as distance measures. Results 
indicate that DataFITS significantly increased road coverage by 137% and improved 
information quality for up to 40% of all roads through data fusion. Traffic estimation 
achieved an R2 score of 0.91 using a polynomial regression model, while incident 
classification achieved 90% accuracy on binary tasks (incident or non-incident) and 
around 80% on classifying three different types of incidents (accident, congestion, 
and non-incident).