Modelling Environment: Evaluating Numeric Expressions

In addition to adding a numeric variable as a dimension to the Modelling Environment, it is possible to right-drag a numeric expression onto the Dimensions panel.

For example, identifying the number of bookings in the last year (requiring an expression) may be more relevant than the total number of bookings (variable). When you add an expression, you can judge the level of insight it provides and decide if it is then worth creating the virtual variable.

At the moment, the actual expression cannot be used with the other modelling tools. Creating the expression and evaluating it in the modelling environment allows you to see if that expression would be insightful in your models. Should you wish to use the expression in a decision rree, for example, you would need to create a virtual variable of the expression.

• Create an expression to identify people who have made bookings in the last year - as shown in the screen shot below:

Having created the dimension, it can now be dragged on to the Dimensions tab of the Modelling Environment.

Let's consider an example further:

• Make a selection of people who have booked a holiday to Greece

  

• Open the Modelling Environment and drag the selection on to the Selections tab as your Analysis Group

• Drag your numeric expression onto the Dimensions tab

In this example, other variables are also added as dimensions:

In the first column, you can see the expression icon, the selector variable icon, and an icon indicating a numeric variable that has been banded. There is also an additional Mean PWE column which allows the dimensions to be ranked.

• Click the blue Build button

The aim of the Modelling Environment is to evaluate the best dimensions and decide which dimensions to use when creating a model.

In this case the expression has the lowest Mean PWE and, as such, you may not want to use it when building your models.

• Click on the Profile tab next to the Associations tab

This column chart allows you to further evaluate the expression. You can switch between dimensions using the drop-down, as well as chose a variety of measures. Here you can see there is very little difference between the Greece Holidaymakers percentage and that of All People; as such, the expression is likely to offer little insight.

In the screen shot below, you can see an example of a user who has been exploring which of the expressions created will be most insightful. In this case, the Mean PWE and Profile both show that the expression ‘Last cost – First cost’ would be useful.

With the evaluation complete, you may wish to delete the other expressions from the Dimensions tab. Should you want to use the ‘Last Cost – First cost’ expression within a particular model, you must first create it as a virtual variable and then drag that virtual variable on to the Modelling Environment.