Our latest release of Modeller, version 5.0, is now available to all new and existing users. We are really excited about the advances brought to our users. The major addition is the “Open source node”. This enables users to run Python scripts and packages directly within Modeller, enabling greater flexibility, efficiencies and ease of use along with the vast amount of choice and data science capabilities that Python offers. Without having to move between tools or platforms it enables the easy sharing of data, assets and models between Modeller and Python. This means the strengths and credit risk domain features of Modeller such as Reject Inference, Grouping, Reporting, Field Reduction can easily be used in combination with Python scripts when needed. The co-ordination is all controlled from within Modeller and it’s in-built audit trail, replicability and documentation provides the strong model governance and model risk management controls required during model development processes.

For example, you can now quickly and easily

- take groupings from Modeller and use within a Python model build
- run your standard suite of Modeller model performance reports against a Python model
- develop a model in Python using a dataset already passed through Modeller’s Field Reducer

and much, much more.

The look and feel of the tool has also been updated in line with our new Paragon branding. We continue to improve and advance Modeller to provide “the best of both worlds”, such that you do not have to make compromises, providing

- Structure with flexibility
- Automation with control
- Traditional logistic regression with machine learning
- In-built algorithms with open source

Modeller is a best-in-class software tool for the development of powerful predictive models with over thirty years of credit risk modelling expertise wrapped in. If you would like to learn more, please get in touch by emailing info@credit-scoring.co.uk