3. True or False? [16] (a) Decision tree models are not robust, and training on different samples can give different models. True (b) Increasing the number of trees in a random forest can lead to overfit. fatu . () Increasing the number of trees in AdaBoost can lead to overfit. Falock (d) A model with higher bias (and low variance) should be preferred over a model that is unbiased. False (c) A bootstrap sample will typically have less than 60% of the examples in the training dataset. True (f) Boosting is an approach to combine multiple strong learners to get better performance. False (g) In using PCA to try reduce the variables to consider for a predictive model, one typically includes the dependent variable for PCA computation. False (h) In SVMs. the penalty factor C can control the tradeoff between model complexity and training error