Near the middle of the chart (2003-2009) we have a peak in traffic enforcement. For several years after that, traffic fatalities declined. Then as enforcement decreased fatalities began to rise.

This paper obtains several theorems giving necessary and sufficient conditions for a directed graph to be a network graph, and applies the results to solving some problems of smoothing traffic flow.

Near the middle of the chart (2003-2009) we have a peak in traffic enforcement. For several years after that, traffic fatalities declined. Then as enforcement decreased fatalities began to rise.

Near the middle of the chart (2003-2009) we have a peak in traffic enforcement. For several years after that, traffic fatalities declined. Then as enforcement decreased fatalities began to rise.

Google today released TensorFlow Graph Neural Networks (TF-GNN) in alpha, a library designed to make it easier to work with graph structured data using TensorFlow, its machine learning framework.

Almost every year prior to 1990 exceeded 750 traffic fatalities. From 1968 (as far back as I could find records) to 1990 we averaged 827 traffic deaths a year, peaking at over 1000 in 1979.