SPIN Processed News Frame: The Fog
Mutual information and sensitivity analysis for feature selection in customer targeting: a comparative study
A comparative study evaluates mutual information and data-based sensitivity analysis for feature selection in bank telemarketing, finding mutual information selects 13 features with slightly better performance at high false positive ratios, while sensitivity analysis selects 9 features and achieves lower false positives.
Spin 20% Claim Present in Source
What AI may repeat
"A study found mutual information selects more features but performs better when false positives are acceptable, while sensitivity analysis uses fewer features and reduces false positives."
arXiv Machine Learning
Aug 24, 2026