Suppose that this routine has an adjustable cutoff (threshold) mechanism by which you can alter the proportion of records classi
fied as fraudulent. Describe how moving the cutoff up or down would affect 2. a. the classification error rate for records that are truly fraudulent (circle best answer) (i)Lowering the cutoff (here, below 0.5) leads to classifying more non-fraudulent records as fraudulent records (more zeros misclassified as 1) (ii)Lowering the cutoff (here, below 0.5) leads to classifying fewer non-fraudulent records as (iii) Lowering the cutoff (here, below 0.5) leads to classifying more fraudulent records as non- (iv) Lowering the cutoff (here, below 0.5) leads to classifying fewer fraudulent records as non- fraudulent records (more zeros misclassified as 1) fraudulent records (more ones misclassified as 0) fraudulent records (more ones misclassified as 0) b. the classification error rate for records that are truly nonfraudulent (circle best answer (i)Increasing the cutoff (here, above 0.5) leads to classifying more non-fraudulent records as fraudulent records (more 1 misclassified as 0) (ii)Increasing the cutoff (here, above 0.5) leads to classifying fewer non-fraudulent records as fraudulent records (more 1 misclassified as 0) (iii)Increasing the cutoff (here, above 0.5) leads to classifying more fraudulent records as non- fraudulent records (more 1 misclassified as 0) (iv) Increasing the cutoff (here, above 0.5) leads to classifying fewer fraudulent records as non- fraudulent records (more 1 misclassified as 0)
a.The classification error rate for records that are truly fraudulent: By increasing the cutoff value, the non-fraudulent records will go down and this increases the error rate.
Also, if the cutoff value is down, the fraudulent records will increase and this lowers the error rate
.b.The classification error rate for records that are truly non-fraudulent: By moving up the cutoff value, the fraudulent records will go down and this decreases the error rate.
Also, if the cutoff value is down, the non-fraudulent records will go down and this increases the error rate.
We have the following values: Total people: 31 People per team: 4 The number of teams will then be: N = (31) / (4) N = 7.75 Round to the previous whole number. N = 7 There are 7 teams. Answer: they can make about 7 teams.
