Applying Short-term Traffic Prediction Models for Updating 
		    Missing Values of Traffic Counts

		 Ming Zhong, Pawan Lingras, and Satish Sharma

The presence of missing values is an important issue for traffic data 
programs. The analyses applied to data sets from two highway agencies 
show that significant portion of data has missing values. Literature 
review indicates that previous research mainly focused on detecting 
missing values. There is limited research on data imputation in 
traffic analysis. In this study, genetically designed neural network 
models and regression models were applied to six permanent traffic 
counts (PTCs) from Alberta, Canada to investigate their merits in 
imputing missing values. These six PTCs belong to different trip-pattern 
groups and functional classes. A top-down model refinement was used to 
search for the models with reasonable accuracy for each type of road. 
Average errors for refined models were lower than 2% and the 95th 
percentile errors were below 4-5% for counts with stable patterns. 
Even for counts with relatively unstable patterns, average errors 
were below 6% in most cases, and the 95th percentile errors were 
rarely more than 10%. It is believed that the models proposed in this 
study would be helpful for highway agencies in their traffic data 
programs.