Fig. 3From: A comparative study of models for the incident duration predictionMLR prediction results versus duration test dataBack to article page