:: Visual Choice ::

Behavioural Choice Modelling Software

:: Navigation
:: Choice Nest Modelling

To build any hierarchy of dependent choices and run your model

This is a uniquely simple, yet powerful way of applying a complex hierarchical nesting of choice models such as are needed for detailed behavioural transport models. You can graphically design your choice nests and work through their complexities to check them and apply them in your transport model. The choice hierarchy can be applied to individual people, agents, activities, tours or trip ends and can produce decisions, probabilities or trip matrices. The software is fast so is capable of swiftly running complex hierarchies for transport models such as the WebTag choice hierarchy of trip frequency, mode, time period, destination, peak spreading and route choices. You can build such a hierarchy in minutes.

A choice hierarchy is composed of a set of individual dependent choices. In Visual Choice, each choice is given a logit choice model which has a set of attributes and their coefficients which are put together in the utility function in order to forecast the probability the chooser decides on one alternative rather than another. The attributes can be skim matrices, zonal planning data, person variables or Alternative Specific Constants.

Different choices which depend upon each other, such as the choice of travel mode and destination, can be connected together in Visual Choice to form a choice hierarchy or choice nest. The nest is operated in the conventional way whereby the logsum or inclusive value from the nest below is passed up to its parent choice for inclusion in the parent's utility function. Logsums are thereby propagated up the nest to the top where the top choice calculates the probability of choosing each alternative. This is passed down to the dependent choices below it to emerge at the bottom of the nest as the probability of each dependent alternative. If trips are applied to the choice nest the output is the trips by each dependent alternative. If required the choice nest can instead use Monte-Carlo micro-simulation to decide on the one chosen alternative at the bottom of the nest. It can use tours (for tour based models), activities, agents or people.

The hierarchy of dependent choice models is given in the Model View pane (figure 1) where clicking on the + sign opens the tree and clicking on the - sign collapses the tree rather like the directories in windows explorer. Choices can be added to the choice nest, deleted or renamed from the main menu. Whole branches of dependent choices can be copied and pasted onto a different choice model to build up your hierarchy and then edited in the Properties pane (see figure 2) which displays the properties for the selected choice model. Alternatively you can edit your choice model in the easy-to-understand wizard (see figure 4).

Insert a Demand Scenario into your model in order to run your choice nest (see figure 3). A Demand Scenario can be a set of trip generations, tours, activities, agents or people - whatever you need for your choice nest. Your choice nest is applied for each row of the demand scenario file. Figure 5 shows a set of tour productions for each market segment so if the choice nest has a destination choice model, it will generate a set of trip matrices.

You can check the logit model calculations by selecting a choice model in the Model View pane and selecting a tour in the Tours tab of the Data pane. Figure 7 shows the calculation for origin zone 4 which displays for each alternative (horizontal axis), the attributes, the coefficients and the calculated utility, exponentiated utility and probabilities for each alternative. You can view the coefficient and attribute files by clicking on the ellipses in the Properties pane (figure 5 shows the coefficient file contents). The calculations can be exported to Microsoft Excel for further analysis by selecting the cells to be exported and going to the Export to Excel option on the Edit menu.

The Model View pane showing the current choice hierarchy. The choice model AMPeak_DestChoice has been selected.

Figure 1 The Model View pane showing the current choice hierarchy. The choice model AMPeak_DestChoice has been selected.

The Model Properties pane showing the properties of the choice model selected in the Model View pane.

Figure 2 The Model Properties pane showing the properties of the choice model selected in the Model View pane.

Add a Demand Scenario to run your model.

Figure 3 Add a Demand Scenario to run your model.

The Choice model wizard can be used to create a choice model; edit the currently selected choice model or edit any other existing choice model more easily than in the Model Properties pane.

Figure 4 The Choice model wizard can be used to create a choice model; edit the currently selected choice model or edit any other existing choice model more easily than in the Model Properties pane.

A Demand Scenario showing the tour productions from each zone for each market segment where each column contains: the origin zone number, two market segment numbers and the number of tours generated from this origin zone.

Figure 5 A Demand Scenario showing the tour productions from each zone for each market segment where each column contains: the origin zone number, two market segment numbers and the number of tours generated from this origin zone.

The Coefficients for the currently selected choice model.

Figure 6 The Coefficients for the currently selected choice model.

The logit calculation of the currently selected tour going through the currently selected choice model. This shows for each alternative (horizontal axis), the attributes, the coefficients and the calculated utility, exponentiated utility and probabilities for each alternative.

Figure 7 The logit calculation of the currently selected tour going through the currently selected choice model. This shows for each alternative (horizontal axis), the attributes, the coefficients and the calculated utility, exponentiated utility and probabilities for each alternative.