1.5 Which of the following specify (or specifies) an over-fitting problem? (10 points) (a). A regression model is more complex than necessary, so although its training error is small, it's the test error is large. (b). A statistical model achieves both a poor training error and a poor test error. (c). You observe the following training and test root mean square (RMS) error curves for an algorithm when you vary the choices of its tuning parameter. You are now using ln =37, so based on the curves, your model over-fits the training data. (d). A statistical model is so simple that it cannot learn the underlying relationship presented in the data. (e). A statistical model fits the training data very hard so it fits the random error or noise in the observed data rather than the underlying relationship of the data