Multicollinearity
Multicollinearity: Strong linear dependence among predictors that destabilises coefficients and attributions in statistical models. Predictions may remain good while interpretability and estimate variance worsen. Regularisation, feature selection and PCA offer different remedies. This perspective connects the definition with practical behaviour in artificial intelligence systems.
In italiano: MulticollinearitàLegge V · Overfitting, bias e collasso dei modelliPredictions may remain good while interpretability and estimate variance worsen. Regularisation, feature selection and PCA offer different remedies.