(a) Compute the GR of each attribute Xi, relative to the class distribution. In the Naive Bayes classifier, remove attributes in the ascending order of GR: first, remove P(Xi|cj) such that Xi has the least GR; second, remove P(Xi′|cj) such that Xi′ has the second least GR,……, until there is only one Xi∗ with the largest GR remaining in the maximand P(cj)P(Xi∗|cj). Observe the change of the accuracy for both Gaussian and KDE (Choose bandwidth σ=10 for KDE).
(b) Compute the IG between each pair of attributes. Describe and explain your observations. Choose an attribute and implement an estimator to predict the value of education num. Explain why you choose this attribute. Enumerate two other examples that an attribute can be used to estimate the other and explain the reason.
Last Completed Projects
| topic title | academic level | Writer | delivered |
|---|