Consider a dataset for the classification task where $10\%$ of the instances

$($examples$)$ belong to the $$positive$$ class and $90\%$ of the instances belong to

the $$negative$$ class. Assume that a data analyst has decided to cope with this

class imbalance by using the cost$-$sensitive learning approach, where different

misclassification costs are assigned to false positives and false negatives.

Answer the following questions about the use of this approach. Write down

any assumption that you make $($if any$).$

$($i$)$ Give a brief definition for each of these two terms: $$false positive$$ and

$$false negative$.$

$[2$ marks$]$

$($ii$)$ Give a numerical value for each of the two misclassification costs, i$.$e$.,$

for a false positive and a false negative.

$[2$ marks$]$

$-7-$

$($iii$)$ Justify, briefly, your choice of misclassification costs in your answer to

item $($ii$).$

$[$