$20$

Numeric $4\mathrm{.}8$ points

Predict the likelihood that an administrator completed the task $($Completed$.$task$=1)$ and create a confusion matrix with a $50\%$ cut$-$off to see the accuracy of the predictions.

Use the predict command to create a new dataset with the predicted values included.

Create a confusion matrix with your prediction variable $>0\mathrm{.}5.$

Use the following code to produce the confusion matrix.

admin.logit$.$pred predict$($admin$.$logit$,$ data$=$admin.df$,$ type$=$"response"$)$

table$($admin$.$df$Completed.task, admin.logit$.$pred $>\mathrm{.}5)$

How many cases are true positives?

Type your answer...

$21$

Numeric $4\mathrm{.}7$ points

What percentage of cases did our model misclassify? $($Use no decimal places and no $\%$ sign.$)$

Type your answer...