Consider Model $3,$ which predicts a continuous score:

Actual Model $30\mathrm{.}60\mathrm{.}8$

$10\mathrm{.}98$

$10\mathrm{.}82$

$00\mathrm{.}78$

$10\mathrm{.}75$

$10\mathrm{.}61$

$00\mathrm{.}53$

$10\mathrm{.}49$

$00\mathrm{.}30$

$00\mathrm{.}21$

$00\mathrm{.}10$

For Model $3,$ compute sensitivity$,$ specificity, precision, and accuracy for two thresholds: $0\mathrm{.}6$ and $0\mathrm{.}8.$

$[$Hint: For each of the thresholds, compare with the Model $3$ predicted scores and identify $1$s and $0$s$.$ Use the standard convention where a score greater than or equal to the threshold is predicted as $1($positive$),$ and a score less than the threshold is predicted as a $0($negative$).]$

Assume the following benefits and costs:

The benefit of identifying a True Positive: $4$

The benefit of identifying a True Negatives: $3$

The cost of a False Positive: $-8$

The cost of a False Negative: $-1$

Answer the following questions.

What is the net benefit of running Model $3$ at a threshold of $0\mathrm{.}6$ and $0\mathrm{.}8,$ respectively?

Group of answer choices

$19,20$

$0\mathrm{.}80,0\mathrm{.}70$

$0\mathrm{.}80,0\mathrm{.}57$

$20,19$