Classification mode

In the first step, you can choose between two classification approaches:

• Supervised Classification

  • Use when you have labeled references

  • Goal: predict target variable for new data

  • Prediction for one or multi-classes

• Unsupervised Classification

  • Use when you may not have labels

  • Goal: discover structure, segments, anomalies

  • Adds: Dimension Reduction before clustering (PCA/UMAP)

Each approach has its own set of algorithms and parameters, which can be configured in the subsequent steps.

It is possible to switch between supervised and unsupervised modes at any time, without losing the specific configuration of the other mode. However, most configuration is common between both modes, such as the preprocessing settings. Changing such common settings will impact both modes and may delete existing models or dimension reduction results, requiring retraining or reconfiguration.