Q Research Software

It should be possible to generate multiple sized portfolios for turf in a single output in one go.

It should be possible to run an Elastic Net Regression. Cf. https://en.wikipedia.org/wiki/Elastic_net_regularization

In-built Kano Analysis creator. See: https://the.datastory.guide/hc/en-us/articles/4634099469199-Overview-of-the-Kano-Model

When conducting a choice model analysis in Displayr, it is possible to create a Simulator that allows for the selection of combinations of attributes to see the utility and to experiment with different scenarios. This function is not available in Q because it’s not possible to create drop-down selectors. However, it should be feasible to output, or at the very least create instructions, for how to replicate the same thing in Excel, for those users that require the provision of a simulator to their clients, and who do not have a Displayr license. feature_request-create_or_export_excel_based_simulator_from_q_choice_model_output [ https://displayr.zendesk.com/agent/tickets/56155|https://displayr.zendesk.com/agent/tickets/56155 ]

Ability to create a Conjoint Simulator in Q - the same way as in Displayr.

There is a format called PMML for Random Forests that can be done with custom R code, but could be a built-in option: [ https://en.wikipedia.org/wiki/Predictive_Model_Markup_Language|https://en.wikipedia.org/wiki/Predictive_Model_Markup_Language|smart-link ] [ https://rdrr.io/cran/pmml/man/pmml.randomForest.html|https://rdrr.io/cran/pmml/man/pmml.randomForest.html|smart-link ]

A customer would like us to add the poLCA (Polytomous Variable Latent Class Analysis) as part of our built in menu system. https://cran.r-project.org/web/packages/poLCA/poLCA.pdf

In the random forest output (Create > Classifier > Random Forest), include an additional column to the right of the "Importance (MeanDecreaseGini)" column containing the % Column Share.

We currently have PCA built-in but this user is after confirmatory and exploratory factor analysis.

When you have a conjoint design and run a Hierarchical Bayes analysis based on "Simulated choices from priors", it's not easy getting a tidied version of the simulated data set with both the alternatives selected and a "Version" variable that links back to the original design. The end result should be an R attribute that stores the simulated data set in the real format that is necessary to run the analysis in Q.