Overfitting occurs when a model fits training data too closely, failing to generalize. Overfitted models contain more parameters than justified by data. Occurs when model selection criteria don't match suitability criteria. Extreme example: model memorizes training data perfectly but fails predictions
Natural monopoly occurs when single firm's production costs are lower than multiple firms'. High infrastructural costs create barriers to entry in industries like utilities. William Baumol defined it as industry where multi-firm production is more costly than monopoly
Boosting creates ensemble models by minimizing errors of previous models. XGBoost is widely used public domain boosting software developed by Chen and Guestrin. Boosting requires more maintenance than bagging, like a powerful but expensive Porsche
Manages regulatory and legal affairs in healthcare, energy, and banking sectors. Ensures product and service compliance with local legislation. Oversees regulatory processes from inception to market release. Acts as interface between businesses and government bodies
Stress testing evaluates institutions' resilience against future financial situations. Helps gauge investment risk and assess asset adequacy. Used across industries to test system reliability and product limits
Overfitting occurs when a model memorizes training data patterns and noise. High variance indicates overfitting, while underfitting shows high bias. Training time and model complexity can cause overfitting. Unclean data and insufficient training data size trigger overfitting