Fraud_​Detection_​Quantile_​Method_​Training

0 ×

This workflow uses the Quantile Method to identify fraudulent transactions to check for numeric outliers in credit card data. The quantile method is a type of classification that is well suited to linearly distributed data. We first read in and process the sample data using normalization, then use a quantile-based approach to remove outliers, labeling them as potential fraud. The model's score can be checked at the end through the 'Scorer' node.

The steps we perform are below:
1. Read Training Data
2. Normalize Data and Remove non-outliers
3. Save Models for deployment
4. Mark Outliers and Score Model

URL: Kaggle Dataset https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud

Nodes

• Rule Engine3 ×
• Model Writer2 ×
• CSV Reader1 ×
• Concatenate1 ×
• Normalizer1 ×
• Numeric Outliers1 ×
• Reference Row Splitter1 ×
• Scorer1 ×

Extensions

• FeatureKNIME Base nodes
• FeatureKNIME Javasnippet
• FeatureKNIME Statistics Nodes

Links

• No links available