Fraud_​Detection_​Isolation_​Forest_​Deployment

0 ×

This workflow reads the trained isolation forest model, as well as the incoming transaction and applies the model to it. Based on the isolation number (mean length) and a threshold, the 'Rule Engine' node detects fraudulent transactions and sends an email, if the isolation number is lower than the specified threshold on Mean Length. The isolation forest model is a part of the KNIME H2O Machine Learning Integration.

This workflow demonstrates how we can use the trained Random Forest Model on new data by performing the following steps:
1. Read the model and new data
2. Apply the model on the new transaction

Nodes

• CASE Switch Start1 ×
• Component Input1 ×
• Component Output1 ×
• H2O Isolation Forest Predictor1 ×
• H2O Local Context1 ×
• Show all 12 nodes

Extensions

• FeatureKNIME Base nodes
• FeatureKNIME H2O Machine Learning Integration
• FeatureKNIME Javasnippet

Links

• No links available