Pareto Optimization¶

Explore accuracy-interpretability trade-offs.

Overview¶

Pareto optimization finds multiple solutions representing different trade-offs between objectives.

Running Pareto Analysis¶

python scripts/run_pareto_optimization.py --config configs/paper.yaml --dataset breast_cancer

Output:

Console: Summary of Pareto solutions
Figure: Pareto front visualization

Understanding Results¶

Example output:

Config   Acc Weight  Test Acc      Interp        Nodes  Depth
1        0.30        0.8823        0.7234        8      3
2        0.50        0.9015        0.6543        11     4
3        0.70        0.9234        0.5123        17     5
4        0.90        0.9456        0.3456        28     6

YOUR CONFIG (★): Acc Weight=0.68, Test Acc=0.9105, Nodes=6.5

Interpretation¶

Config 1: High interpretability, lower accuracy
Config 2-3: Balanced solutions (★ your config)
Config 4: High accuracy, lower interpretability

Choose based on your domain requirements!

See Multi-Objective Guide for implementation details.