Skip to main content

Model Evaluation

Comprehensive model evaluation using cross-validation, multiple metrics, confusion matrices, ROC/PR curves, and performance comparison against baselines.

Complexity: Moderate | Duration: 15-30m | Category: Devops

Tags: ml evaluation validation metrics

Workflow Diagram​

Steps​

Step 1: Cross​

Agent: validation

ml-engineer - K-fold CV with stratification

Step 2: Metrics computation​

Agent: ml

engineer - Accuracy, precision, recall, F1, AUC-ROC, AUC-PR

Step 3: Confusion matrix​

Agent: ml

engineer - Analyze TP, FP, TN, FN distributions

Step 4: ROC/PR curves​

Agent: ml

engineer - Plot and analyze curve shapes

Step 5: Baseline comparison​

Agent: ml

engineer - Compare against dummy classifier/regressor

Step 6: Error analysis​

Agent: testing

specialist - Identify patterns in misclassifications

Step 7: Performance report​

Agent: testing

specialist - Generate comprehensive evaluation report

Step 8: Acceptance review​

Agent: testing

specialist - Validate against business requirements

Usage​

To execute this workflow:

/workflow devops/model-evaluation.workflow

See other workflows in this category for related automation patterns.