This procedure will replicate the bicycle route choice of a user. It performs a HIGHWAY assignment twice; once for skimming and once for assignment. In addition, it would be suitable to select the proper variables in the generalized cost functions as well as calibrate the coefficients in the utility functions.

Step 1: Run Highway for only one iteration to get skimming values

• Using the generalized cost functions including any available components related to the bicycle trips
• Use a suitable number of multiple PATHLOADs (e.g. 3 or 4 – depending on the are) with different path costs to replicate the stochastic route choice
• Skim outputs to include all the route information such as number of junctions

Step 2: Run Choice in Matrix to compute the probability for each route using discrete choice model

• Compute utility values for each O-D pair using the skimming outputs for routes
• Apply a multinomial logit model (using Choice in matrix) to estimate the probabilities for available routes

Step 3: Apply probabilities to split the original O-D trips to each route

• This step eventually generates the multiple O-D trips per each O-D pair

Step 4: Run Highway for only one iteration to assign the bicycle trips for each bicycle route

• Should use the same generalized cost functions used in Step 1
• Should use the same number of PATHLOADs used in Step 1