Quiz13

A disadvantage of kNN is: drill

A disadvantage of kNN is:

Answer

  • ☑ time complexity to predict
  • ☐ time complexity to train + [ ] space complexity to train
  • ☐ space complexity to predict

For bayes classifier drill

For bayes classifier

Answer

  • ☑ estimating \(p(x|c_i)\) is hard
  • ☐ estimating \(p(c_i)\) is hard
  • ☐ The assumption of independence is hard
  • ☐ dealing with numerical data is hard

Naive bayes assumes that

Naive bayes assumes that

Answer

  • ☐ \(p(x_1,\dots,x_d) = p(x_1) . \ldots . p(x_d)\)
  • ☑ \(p(x_1,\dots,x_d | c_i) = p(x_1|c_i) . \ldots . p(x_d | c_i)\)
  • ☐ \(p(c_i | x_1,\dots,x_d) = p(c_i| x_1) . \ldots . p(c_i | x_d)\)
  • ☐ \(p(x_1,\dots,x_d) = p(x_1|c_i) . \ldots . p(x_d | c_i)\)

In the context of Naive Bayes classifier, Laplace correction is used to address

In the context of Naive Bayes classifier, Laplace correction is used to address

Answer

  • ☐ Ensuring that \(p(x_j = t | c_i) = 0\) when there are no points with \(x_j = t\) for all data points of \(c_i\)
  • ☐ Ensuring that \(p(x_j= t) = 0\) even when there are no points with \(x_j = t\) for all data points of \(c_i\)
  • ☑ Ensuring that \(p(x_j= t | c_i) \neq 0\) even when there are no points with \(x_j = t\) for all data points of \(c_i\)
  • ☐ Ensuring that \(p(c_i) \neq 0\) no data point belongs to class \(c_i\)