Question: In one to three sentences, explain why reducing or removing regularization can decrease bias while increasing variance.

Sample answer: Regularization restricts how freely a model can adapt to the training set. If that restriction is weakened, the model can fit the training examples more closely, which can reduce systematic error, but it may also start tracking random quirks in the data, making its predictions less stable on new examples.

Key points:

• Regularization constrains model flexibility
• Less regularization allows a tighter fit to the training set
• Tighter fitting can reduce bias
• Tighter fitting can also make the model more sensitive to noise, increasing variance

Rubric: Full credit for stating that regularization limits flexibility or fitting power, and that reducing it can improve training fit and lower bias while increasing sensitivity to noise or overfitting, which raises variance.