Research Article

An Improved Deep Learning Model for Traffic Crash Prediction

Table 1

Summary statistics of analyzed continuous variables.

Variable MeanStd. Dev.Min.Max.

Independent variable
The number of major injury crashes per year per roadway segment0.040.3203
The number of minor injury crashes per year per roadway segment0.410.5409
The number of no injury crashes per year per roadway segment1.231.74022
Traffic factors
The logarithm of AADT per lane3.533.742.934.51
Truck traffic percentage6.714.98133
Posted speed limits38.656.693070
Geometric design features
Segment length (miles)0.811.030.0212.31
Degree of horizontal curvature1.513.23014.00
Median widths1.122.02012
Outside shoulder widths3.061.883.528
Pavement factors
International roughness index65.8527.7525.45182.58
Rut depth (in.)0.130.050.060.41