# Machine learning short answer questions, machine learning university exam questions

__Machine learning questions__

*1. In classification, what does high entropy mean? *

*2. Which clustering approach is the best to produce clusters of different sizes and shapes? **
*

*3. Which type of machine learning algorithms would be helpful to predict amount of rainfall in a region? *

*4. In linear regression,which regularization penalties can be used to reduce some parameters to zero? **
*

*5. Why the MLE estimates are often considered as undesirable? *

*6. If you have enormous amount of training data, what would be the variance of your model trained on that data? *

*7. In a neural network, if you have more neurons in the hidden layers, what would be the impact? *

*8. What will be the problem with the decision tree that is too shallow?**
*

*9. How do we refer to a machine learning model that neither model the training data nor generalize to new unseen data? *

*10. When can hard margin SVM work?*

*11. What is the objective of K-means clustering?*

*12. How does polynomial degree affect the overfitting and underfitting in polynomial regression?*

*13. When does bias become low or high in polynomial regression?*

*14. How does the use of weak classifiers help preventing overfitting when perform bagging?*

*15. How does “the averaging the output of multiple decision tree” help?*

*16. What are some assumptions made by k-means algorithm?*

*17. What is an activation function in neural networks?*

*18. How does the kernel width affect the trade-off between underfitting and overfitting in kernel regression?*

*19. Why do we need to prune a decision tree?*

*20. Discuss on bias and variance tradeoff.*

*21. What is the cause of increase in training loss with number of epochs?*

*22. Why does decision tree
algorithm can achieve zero training error on any linearly separable dataset?** *

*23. Why does perceptron can achieve zero training error on any linearly separable dataset?**
*

*24. List down different ways to reduce the overfitting problem.*

*25. Which is the most suitable error function for gradient descent using logistic regression?*

*26. When do we use linear kernel while training an SVM?**
*

*27. What is the impact of increasing the size of layers in a neural network on bias and variance?*

*28. How does the increase in the number of hidden units per layer in a neural network affect bias and variance?*

*29. How does pruning help in the development of a decision tree?*

*30. What is the major weakness of decision trees when compared with logistic regression classifiers?*

*31. Why decision tree is not an ensemble machine learning method? *

*32. What is sequential ensemble?*

*33. Differentiate between sequential and parallel ensemble models.*

*
*

#### 34. *What will be the effect of k-nearest neighbor model on bias and variance if we increase the value of k?*

*What will be the effect of k-nearest neighbor model on bias and variance if we increase the value of k?*

#### 35. *What would be the bias-variance tradeoff if we increase the value of k in k-nearest neighbor?*

#### 36. *Will k-nearest neighbor model underfit or overfit for a large value of k?*

#### 37. *Differentiate between underfit and overfit in machine learning*.

#### 38. *What is the impact of multi-way split in decision tree learning?*

* *

* *

* *

* *

* *

* *

*****************

**Related links: **

*Machine learning MCQ home page**Machine learning True / False questions home page**Machine learning exam questions, machine learning exam questions, machine learning questions and answers, machine learning questions and answers, machine learning university exam, machine learning university exam*