Advanced data analysis (EXAMEN van Bo !! 2018-2019):
1, 2, 3 en 4: calculations pertaining to true positives, recall, precision…
Check if a hierarchical cluster with complete linkage was correct, if you compared them with the distances
between the datapoints -> they gave 4 options, fifth option was that none were correct
What makes deep learning so good? (MCQ):
a) they can account for complexity
b) they have hidden layers
c) …
d) ..
They showed an ROC curve and a PR curve with points in the corners and one in the middle, with
corresponding letters:
→ if it is random, what points would each curve go through?
→ if every sample was placed in the wrong class, what points would the curves go through?
→ mcq about what you could say with absolute certainty when looking at the curves
a) it belongs to a neural network
b) the sensitivity/specificity trade off is shit
c) there are more positive than negative values
d) there are more negative than positive values
e) None of the above
1, 2, 3 en 4: calculations pertaining to true positives, recall, precision…
Check if a hierarchical cluster with complete linkage was correct, if you compared them with the distances
between the datapoints -> they gave 4 options, fifth option was that none were correct
What makes deep learning so good? (MCQ):
a) they can account for complexity
b) they have hidden layers
c) …
d) ..
They showed an ROC curve and a PR curve with points in the corners and one in the middle, with
corresponding letters:
→ if it is random, what points would each curve go through?
→ if every sample was placed in the wrong class, what points would the curves go through?
→ mcq about what you could say with absolute certainty when looking at the curves
a) it belongs to a neural network
b) the sensitivity/specificity trade off is shit
c) there are more positive than negative values
d) there are more negative than positive values
e) None of the above
