pathfinding probability
The probability of each path in the pathset based on the pathfinding cost. Differs from simulated probability in that the simulation step hasn’t occurred yet so the probabilities are subject to change; for example, some paths may become infeasible if the simulation fills up some vehicles so the the passenger cannot board the vehicle for this leg. May also include path overlap penalties (although Fast-Trips doesn’t calculate it in Pathfinding, only in Simulation).
Related Terms
acyclic subgraph
An acyclic subgraph is a network without cycles (a cycle is a complete circuit). When following the network from node to node, a node is never visited twice.
adjacent notes
Nodes that are reachable from a node via a single link
arrival time window
A timeframe within which all paths that the passenger considers must arrive at the destination
backward shortest path
A the shortest path computation that starts at the destination node. This is used when there is a preferred arrival time.
departure time window
A timeframe within which all paths that passenger considers must depart the origin.
downstream link
A link that is reachable from a node (forward shortest-path) or a link that can reach the current node (backward shortest path)
downstream node
A node that is reachable from a node (forward shortest-path) or a node that can reach the current node (backward shortest path)
emanating link
Links coming into or going out of a node.
hyperlink
A hyperlink consists of a set of transit links from a single origin to multiple possible destination nodes [ for forward-shortest-path finding ] or a set of links from a single destination to multiple possible origin nodes [ for backward-shortest-path finding ]. There may be multiple transit departures within a hyperlink.
hyperpath cost
The disutility of a specific hyperpath or portion thereof for a specific user class. Based on both link features, path features, and the path weight parameters.
hyperpath
An acyclic subnetwork with at least one link connecting the origin to the destination, and where at each node, there are probabilities for choosing the alternative links. In most hyperpath-based frameworks, this can be equivalent to the path choice set.
kirchoff model
An alternative formulation to the path-size logit model which considers the relative size of the total utilities of the path. For example, deltas of short paths are perceived differently than deltas on long paths. For paths that are generally the same size, the results will be similar to a logit model formulation.
departure time window
The ordered queue that contains stops with stop states that need to have their costs updated in order to find the shortest path. Stops are processed in order of cost, with least cost stops processed first.
link feature
A feature that can be accurately represented for an individual link (e.g. in-vehicle-time, wait-time, walk-time). Some features, like fares, can be represented at a link level, but may not be accurate if used without understanding of the entire path due to dependent features such as maximum fares or transfers.
path choice model
A mathematical model for selecting a path from a set of feasible choices. Examples include the recursive logit model, path-size logit model, Kirchhoff model.
path feature
Feature of a path that is summarized for the entire path. Can be an aggregation of the link features comprising the path (i.e. wait-time), or a variable that can only be calculated at the level of the entire path (e.g. fare, directness).
path overlap variable
An additive measure of similarity between paths. In road-based path-choice models, this is often the distance of the shared links. In passenger-based path-choice models, this could include considerations for shared-routes, similar-routes, on/off stations, traversed stations, traverse distance and more. If the overlap variable is an indicator variable (𝛿), then it can be 1 or 0; if it distance or cost, then it is a continuous variable.
path overlap
In a set of possible paths through a transportation network, some portion of each of the paths may share a facility meaning that each choice in a choice set is not mutually exclusive. This is important in the context of choice modeling, since it violates the “Independence” in the IIA property of a Multinomial Logit. Formulations that compensate for this violation by discounting the “independence” of each path based on a measurement of commonality (the path overlap variable) include the path-size logit model.
path weight parameters
Coefficients on various components of the utility equation for the path choice model. Can be estimated, asserted, or inferred. Path weight parameters can either be based on a link feature (i.e. in-vehicle time) or path feature (i.e. directness). The path weight parameters can vary by user class.
pathfinding cost
The cost of each path as calculated during pathfinding, typically including costs of each leg plus some non-additive costs. Because of the non-additive components (e.g. fares) and the adjustment of some components (e.g. travelers may leave their origin later than their preferred time to match their bus departure time), it’s typically computed fully only after the path is fully defined (origin to destination and all legs in between). Differs from the simulation cost in that the travel times may not be accurate because the simulation step hasn’t happened yet which would load passengers onto vehicles, thereby slowing them down for boarding/alighting and possibly bumping passengers.
joint person tour
A set of paths between an origin and destination with specific costs, waypoints, and timings. In hyperpath-based frameworks this is often derived from the hyperpath.
predicted passenger path choice set links
A list of links for a predicted path set/hyperpath, with their timing and costs, that the passenger considers taking.
predicted passenger path choice set
A list of feasible paths, with their timing and costs, that the passenger considers taking for a specific trip. In some frameworks, hyperpaths can be used to create the path choice set.
simulated cost
The cost of each path as calculated during simulation. It differs from the pathfinding cost in that it includes the effect of simulation (loaded vehicles and updated vehicle travel times, crowding costs or costs associated from being bumped from the vehicle).
simulated probability
The probability of each path as calculated during simulation. Like the simulated cost, it differs from the pathfinding probability in that it includes the effect of simulation (loaded vehicles and updated vehicle travel times, crowding costs or costs associated from being bumped from the vehicle, which may make the path infeasible, giving it 0% probability).
upstream link
A link that is reachable from a node (backward shortest-path) or a link that can reach the current node (forward shortest path)
upstream node
A node that is reachable from a node (backward shortest-path) or a node that can reach the current node (forward shortest path)
user classes
Passenger market segments who use the same parameters (including path weight parameters) to find and assess paths. Common user class segmentations are by travel purpose and captive versus choice users.
Relevant Articles