- Landing page: this page allows you to provide the name of the project along with key information, such as project number, fieldwork information, and your team contact details. You can include your client’s logo as well as images/ descriptions of the segments that the tool can identify.
- Allocation model: the tool can allocate respondents based on either a linear or a quadratic discriminant model, depending on the model specified within the R-sw Discriminant function discriminant.analysis. This function returns a number of metrics and these need to be pasted into the allocation tool.
- Predictions: the tool can allocate one respondent at-a-time or multiple respondents in one go, depending on the requirements.
- All possible combinations: the tool is also able to produce a list of all possible combinations of all values of the variables and, for each combination, the most likely segment is provided. This feature is particularly helpful when the segment allocation needs to be carried out on a separate IT platform and the developer, rather than implementing a complex allocation algorithm, prefers obtaining the most likely segment for a given record by looking into a precompiled list of all possible combinations of variables with the associated segment. Currently, a maximum of 1,048,574 combinations can be considered (due to limitations of an Excel worksheet), but changes in the tool macros could allow a larger number of combinations by splitting them onto multiple Excel worksheets. In order to keep the number of combinations down at manageable levels, the tool automatically adopts a binning approach for continuous variables.
- Limitations: as it is, the tool can handle up to 5,000 records, 30 segmentation variables and 10 segments; however, these limitations can easily be overcome by changing some of the tool macros (Demia can help if required).
Step-by-Step guide to amend the Allocation Tool
- Review the landing page (worksheet mainOutput) and add the team contact details for the project and your/your client logo;
- load into the simulator the data (worksheet Data), the segment information (worksheet MultiPrediction) and the model metrics (worksheets Metrics and ScalingQuadratic);
- click on the button on the worksheet MultiPrediction to identify the most likely segment for all records in the data;
- check predictions rates reported on the worksheet PredictionChecks by clicking on the button at the top;
- prepare the graphical interface by adding the relevant scales (available in the worksheet Library) to the worksheet ToolOutput and map them to the correct segmentation variables. Detailed instructions on how to use the R-sw Allocation Tool are sent with the tool. If you need another copy, please get in touch with us firstname.lastname@example.org;
- review the help page by replacing the screenshot and by adapting the explanatory notes;
- change the simulator graphics (buttons’ colors and dimensions, background color, font type and size, etc.) as required.