Besedilo vprašanja

Which statements about Bagging (Bootstrap Aggregating) are correct?

Izberite en ali več odgovorov:

Bagging is prone to overfitting

The main idea is to apply a single algorithm to different samples of the original data

The main idea is to apply different algorithms to the same data

Bagging improves the performance of algorithms with high variance

Besedilo vprašanja

A classifier was trained using a learning set that contains 200 examples: 20 from the class C1, 76 from the class C2, 37 from the class C3, and 67 from the class C4. The classifier was tested on an independent set of examples. The results are summarized in the following confusion matrix:

                        \ Predicted class
Observed class \		 C1 C2 C3 C4
C1 14 3 4 4
C2 1 13 0 2
C3 5 0 16 15
C4 16 8 2 47

Which of the below statements are correct?

Izberite en ali več odgovorov:

The obtained average Brier score of the model is 0.82.

The obtained classification accuracy of the majority classifier is 0.4533.

The obtained classification accuracy of the majority classifier is 0.107.

The obtained classification accuracy of the majority classifier is 0.24.

The obtained average Brier score of the model is 1.2456.

The average Brier score of the model cannot be calculated from the given confusion matrix.

The obtained classification accuracy of the model is 0.9.

The obtained classification accuracy of the model is 0.6.

Besedilo vprašanja

Given is a data set consisting of 1000 examples (600 from the positive class and 400 from the negative class. What happens if we use a k-nearest neighbors model by setting the parameter k to the number of training examples?
(select one or more answers)

Izberite en ali več odgovorov:

this model is equivalent to the majority classifier

the specificity of that model is 1

the classification accuracy of that model is 0

the classification accuracy of that model is 1

the sensitivity of that model is 1

Besedilo vprašanja

The classifier was tested on a two class problem and we obtained the following confusion matrix:


      \  Classified as 
True \
class \		 0 1
0 250 50
1 90 110

Which measures can be calculated using the confusion matrix?

Izberite en ali več odgovorov:

classification accuracy

sensitivity

specificity

AUC (area under the ROC curve)

Brier score

Besedilo vprašanja

We trained a model to predict a numeric target variable. Which measures can be used to evaluate the quality of the model?

Izberite en ali več odgovorov:

information score

sensitivity

mean squared error

classification accuracy

Brier score

specificity

mean absolute error