After creating an optimization scenario, the next step is defining how it will work. This document provides information on the steps and details for defining a scenario.
It is helpful to understand the basic concepts related to doing optimization in LityxIQ. See this document as a starting point: https://support.lityxiq.com/971326-Conceptual-Overview-of-LityxIQ-Optimize.
First, select the scenario from the list of available scenarios in the selected Optimize library. Then click Edit Settings from the Selected Scenario dropdown list. If you just created a new scenario, this step will not be necessary.
The Scenario Settings dialog has a number of tabs, each with a number of options. These are described below:
Dataset & Settings Tab
Scenario Definition Method - The standard setting is "Do Not Use a Template". This setting implies that you will define everything about the scenario directly in this dialog. You also are able to select other scenarios that have been defined in the project. If you select another scenario, that scenario's definition will be imported, and you can always start again with its definition by clicking the Reset Scenario to Match Template button.
Dataset - Select the dataset that will serve as the basis for the data that will drive the optimization results. Any dataset you have permissions for in Data Manager will be available.
Optimization Dimensions - Select the variables that serve as optimization dimensions. The optimization decisions that are made will be made for all unique combinations of the levels of the selected dimensions. A typical example is to select a unique individual ID variable when optimizing decisions for prospects (for example). In this case, the output of the optimization will give a decision such as "target" or "not target" for each unique ID. Mathematically, the combination of the levels of all selected dimensions would be referred to as "decision variables".
Attributes - Select the variables that can be used as a basis for constraints or for reporting. Only string and integer variables (with not too many unique values) can be selected. Although you can select any number of variables in this list, it is more efficient to just select the ones you will actually use since LityxIQ will track data for all selected Attributes.
Data Elements - Select the variables that can you wish to use as quantitative data about each dimension level. They are used to define optimization metrics, and can also be used within constraint definitions to provide dynamic constraint bounds. Only numeric variables are available in this list. Although you can select any or all numeric variables from the dataset in this list, it is more efficient to just select the ones you will actually use since LityxIQ will track data for all selected Data Elements.
Defined Metrics - Metrics are summaries of the dataset than can be used to optimize against, or for reporting purposes. Mathematically, these may be referred to as "objective functions". More information for defining optimization metrics and working with this list is provided in this document: https://support.lityxiq.com/773516-Define-Optimization-Metrics.
On the Objective tab, you will define settings related to how the optimization reaches an optimal result.
Max/Min - Select either Maximize or Minimize. The correct choice depends completely on the Objective Function you pick in the next option, and the overall business objective of your situation.
Objective Function - The metrics that have been defined on the "Dataset & Settings" tab will appear in this list. Only metrics that meet the definition of a mathematical "objective function" will be on the list. Select the metric that you want to optimize (in combination with the Max/Min selection).
Output Label - Enter into this text box the label you would like to assign to the result of this optimization run. For each unique level of the combination of Optimization Dimensions, the optimization run will produce an optimal result in any output datasets. This setting for Output Label will be the name of that variable in the resulting datasets.
Decision Type - Choose the option that matches your problem and the required output.
- Binary - the result of the optimization will be a 0/1 decision (binary decision) for each unique level of the combination of Optimization Dimensions. This is a common setting, and is appropriate for situation where you want to optimize the decision of "doing something" or "not doing something". An example is the decision to target a specific prospect, or not target them.
- Integer - the result of the optimization will be an integer result, for each unique level of the combination of Optimization Dimensions. This setting would be used in situation where you are looking to decide the optimal number of times to do something. For example, the optimal number of emails to send over a certain time period would be a use case for the "Integer" setting.
- Continuous Values - the result of the optimization can be any continuous (real) number. This setting would be used in a situation where the result you require for each unique level of the combination of Optimization Dimensions is a value such a budget level, or perhaps a value such as lifetime value.
The Constraints tab provides a list of all constraints, as well as the ability to add new constraints, or delete and modify existing constraints. This functionality is explained further in this document: https://support.lityxiq.com/845223-Manage-Optimization-Constraints.
The filter tab can be used to specify a subset of the data that will be used in the optimization. See https://support.lityxiq.com/806706-Using-the-Filter-Dialog for more information on using the filtering and searching dialog.
When finished, click Save to save the scenario settings and close the dialog.