Build on the insights learnt through data discovery by using supervised and unsupervised modelling techniques.
Workflow and coding options
Model in the workflow environment with visual tools or choose to write code to take advantage of any modelling algorithms available in Python, R, and from our own SAS language capability.
Whether segmenting markets, identifying customer behaviour or classifying fraudulent behaviour, use techniques including k-means and hierarchical clustering.
Workflow’s drag-and-drop clustering blocks provide quick insights and convenient modelling. Explore results with visuals at the click of a button.
Use our decision forest capability as a supervised modelling method with a reliable predictive performance for continuous and discrete variables.
Select from a range of configuration options to choose the growth algorithm, variable treatment and other preferences that give easy and complete control without the need for extensive modelling experience.
Workflow makes it easy to use neural networks for predictive modelling. Just add a neural network block to a workflow and change any model inputs and configuration options to explore statistical results immediately.
Write linear and logistic regression syntax in your code to generate output or simply add regression blocks to a workflow for instant access to reports with easy exploration of a range of statistical outputs. Change any of the inputs of a workflow regression block to generate real-time updates to the model report.
Create credit risk scorecards using workflow blocks to help visualise the process:
- A weight of evidence transformation block provides tables and graphs to guide binning
- Optimise variables automatically with a range of configuration options, or manually control to join and manipulate bins
- Models are built with an array of statistics and charts and scorecard blocks allow you to scale the results
The model analyser is a workflow-only block that provides instant assessment and comparison of different models to determine the best one for your needs:
- Compare models from train and test partitions
- Compare output from different types of model
- Compare models generated from workflow modelling blocks to those generated from code blocks programmed in Python, R or SAS language
- Any changes to input models are automatically reflected in the model analysis report