False. Model simplification decreases the variance but increases the bias.
Card 1 of 259. Answer: False. Model simplification decreases the variance but increases the bias.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
False. Simplifying a model does the opposite: it increases bias and decreases variance. The statement's conclusion is right, since a model simplified too far underfits, but the cause is the extra bias. A simple model makes strong assumptions, so it misses real structure, yet changes little when trained on a different sample.
Example: prune a decision tree from depth 20 to depth 2. The deep tree memorizes its training set and changes shape with every new sample: low bias, high variance. The depth-2 tree has at most four leaves, so it misses real structure but looks almost the same across samples: high bias, low variance. Stronger regularization or dropping features moves a model the same way.
That is why simplification is the standard fix for overfitting: it trades a little bias for a large cut in variance, and test error falls as long as variance was the bigger problem. A rule of thumb for interviews: high training error points to bias, while a large train-validation gap points to variance.
