Modeller: powerful predictive models
Over thirty years of credit risk modelling expertise wrapped into model building software for today’s age of machine learning.
Modeller is a feature-rich, flexible, interactive and transparent tool that helps organisations get the best from their analytical teams. It supports a choice of techniques, the rapid development of powerful models, full explainability and the advancement of less experienced team members.
Business-led, flexible software
Rapid model development
Fully explainable machine learning
Credit risk specific features
Easy implementation and full support
Data management, manipulation and classification
Import data from a wide range of sources, including SAS, SQL and Excel. The dynamic Data Viewer helps you to understand the data – including univariate analysis, volumes and distributions.
Manipulate the data, create your own characteristics, and define populations, weights, targets and holdout samples.
Use the automated Field Reduction tool to reduce the dimensionality of your data - for storage or sourcing purposes, to improve customer or operational processes, or improve modelling efficiency, by removing variables that have little or no discriminatory power.
‘Autogroup’ makes bucketing quick and easy for statistically robust models and a sound starting point for more in-depth group manipulation, via the point-and-click interface.
During editing, the information for the group or variable, e.g. Weight of Evidence (WoE) or Gini, are updated in real time.
How it works: grouping
Model building, validation and deployment
Choose from numerous modelling techniques, including machine learning, for optimal predictive accuracy – especially on datasets with multicollinearity and complex interrelationships.
Create industry-standard continuous and binary target models at the click of a button.
Use decision tree modelling with CART and CHAID trees. Choose from logistic regression, elastic net models, survival analysis (Cox PH), random forests, XGBoost, stochastic gradient descent and more.
SHAP values explain the output of complex machine learning models.
Modeller’s advanced reporting tool allows you to validate scorecards, compare models developed using different algorithms and techniques, and describe their benefits.
Easy documentation and deployment with multiple options for each.
Modeller comes with an inline tool for real-time documentation and a Microsoft Office plug-in for the documentation of the entire scorecard. It is also compatible with Jupyter notebooks and includes full coding instructions for deployment.
Export options for implementation in other scoring and decisioning software include SAS, SQL, PMML and Python.
With advanced decision engines, such as Paris, models can be deployed without IT involvement.
How it works: modelling
Implementation and Support
Install on your own IT infrastructure or private cloud, or run as a service from public cloud servers.
Adaptable to the requirements of many different types of user, with a point and click interface that couldn’t be easier.
Low and no code options make getting the team up and running fast and efficient, as well as ensuring modellers keep to standardised processes.
Integrated management means anyone can arrange everything from workflows to models, reports and more, with a completely transparent development process for ease of oversight and auditability.
Whether you need support with implementation, functionality, or scorecard development, we’re always on hand to help, either directly or through our trusted partners.
Book a demo
Find out today if Modeller could be for you.