Model Dragon 2.0
The no-code platform for rigorous, regulatory-compliant credit risk and fraud model development. Build models, scorecards, decision strategies, and fraud rules — all in one integrated environment.Model Dragon 2.0 brings together everything a credit risk or fraud team needs — from raw data to production-ready models and decisioning strategies. It embeds 25+ years of industry best practices so your team builds faster, with fewer errors, and full regulatory confidence. Available as a standalone server or as a native Databricks App.
Data Integration
Upload CSV files or connect directly to Snowflake and Databricks tables. No ETL pipelines or engineering tickets required. Data is instantly viewable and ready for transformation.
Feature Engineering
Build complex transformation pipelines with point-and-click. Group-by aggregations, merges, derived variables, and BAD flag creation — all without writing a single line of code.
Model Development
Logistic Scorecard, Random Forest, and LightGBM — the most robust techniques for credit risk. Set special values, define variable roles, and build train/test validated models in minutes.
Exploratory Analysis
Descriptive statistics, correlation heatmaps, and bi-variate analysis tables — all available with simple point and click. Understand your data before you model it.
Strategy & Segmentation
Multi-target decision trees for credit approval, limit assignment, and pricing strategy. Define segments using risk scores, bureau attributes, and income — and optimise across multiple objectives simultaneously.
Anomaly Detection
Unsupervised anomaly detection and automated fraud rules generation. Proactively identify and respond to emergent fraud patterns without waiting for labelled data.
Group-by Aggregation
Build a summarised performance dataset from 24-month transaction data using the no-code transformation pipeline.
Merging Datasets
Join application and aggregated performance data to create a modeling-ready dataset — no SQL required.
Credit Strategy Segmentation
Design a multi-target decision tree for credit approval, limit setting, and APR optimisation using risk scores and bureau attributes.
Connect your data
Upload CSV files or connect directly to Snowflake or Databricks. The data viewer loads instantly so you can verify quality before proceeding.
Engineer your features
Use the transformation pipeline to aggregate, merge, and derive variables. Create BAD flags, time-boxed performance summaries, and application-level features — all with drag-and-drop.
Explore and validate
Run descriptive statistics, correlation heatmaps, and bi-variate analysis to understand variable behaviour before modeling. Split into train and test with a single click.
Build and validate your model
Choose from Logistic Scorecard, Random Forest, or LightGBM. Set variable roles, configure special values, and build a validated model with train/test performance metrics.
Design your strategy
Use the Segmentation tree to build multi-target decision strategies — approve/reject, limit assignment, and pricing — using your model score and bureau attributes.
Deploy via Decision Dragon
Export your model and strategy to Decision Dragon for real-time scoring and policy execution in production.
Native Databricks App
Deploy directly on your Databricks workspace. Centralised user management and serverless compute out of the box.
Snowflake Connected
Query tables directly from Snowflake. No data movement, no duplication — work on live data in your warehouse.
Standalone Server
Deploy on your own cloud VM or on-premise server. Full control over infrastructure and data residency.
SR 11-7 Compliant
Built with regulatory model risk management requirements embedded. Explainability, validation, and documentation built in.
- ✓ Up to 30 named users
- ✓ All capabilities included — no module fees
- ✓ Standalone server or Databricks App deployment
- ✓ Snowflake and Databricks connectivity
- ✓ Onboarding support and QuickStart guide
- ✓ Most cost-effective platform in class
No commitment required. We will set up a proof-of-concept on your data.
Talk to us about a free proof-of-concept on cloud. No commitment required.