SPE3DLAB can perform an unsupervised selection of predictors. This allows the identification of the leanest model out from a dataset with a lot of variables.
SPE3DLAB allows you to explore your data in an interactive 3D space. The sample is reduced using a t-distributed stochastic neighbor embedding (t-SNE) algorithm and plotted in an interactive map where the user can select individuals and groups to display their characteristics.
SPE3DLAB allows you to find the best number of clusters based on consensus clustering techniques.
You can then decide the number of clusters to use for the final clustering. The clustering can be analyzed and saved as an additional variable available in all analyses.
Continue reading “Creation of semantic variables based on the document-term matrix (DTM)”