Praexia (formerly Constrata Core Credit)

Praexia is a library that assists with interactive credit modelling in Python.

Welcome to the official documentation for Praexia! If you’d like to dive in immediately with examples, please see the Getting Started section.

For more information on Praexia, and how to obtain a license or instance, please see the official Praexia website.

The modules available in this library can roughly be grouped into:

• data preprocessing (data_access, preprocessing, variable_reduction)

• model development (rollrates, weights_of_evidence_widgets)

• monitoring and reporting (monitoring, monitoring_plots, report)

This documentation contains a full description of all user-accessible classes and functions in the Praexia library.

Notable merits of the library are:

• Practical: Caters for a number of typical processes used for credit model development.

• Interactive: Allows for users to interact with outputs, allowing for rapid feedback and manipulation.

• Flexible: Handles a mixture of data variable types. Additionally, there are options for automation or manual manipulation in certain modules.

On a high-level, this library takes a dataframe of historical default data, as the example shown below.

After processing the data (such as that shown above), users can interact with widgets and develop a probability of default model. This credit model takes data (such as that shown above) and predicts the probability of default.