Original practice · fully worked
Original variant: overnight processing features
For an overnight data-processing batch, event A means that edge compression is used and event B means that redundant storage is used. Historical records give P(A)=0.55 and P(B)=0.35, but they do not specify the dependence between the events. Determine the full possible range of the probability that exactly one of the two features is used.
- A
0.000 to 0.350 - B
0.200 to 0.550 - C
0.200 to 0.900 - D
0.550 to 0.900 - E
0.650
Answer and calculation
If q is the probability that both features are used, the exactly-one probability is 0.90-2q. Since q can range from 0 to 0.35, the answer ranges from 0.20 to 0.90, selecting choice C.
Setup
Setup
Represent the unspecified dependence by the intersection probability q.
q=Pr(A∩B) 0≤q≤min(0.55,0.35)=0.35 Model
Model
Exactly one feature is used in either A without B or B without A. Each marginal contains the intersection once.
Pr(exactly one)=Pr(A)+Pr(B)−2Pr(A∩B) Pr(exactly one)=0.90−2q Compute
Compute
The expression decreases with q, so its endpoints occur at the two feasible endpoint overlaps.
q=0⟹Pr(exactly one)=0.90 q=0.35⟹Pr(exactly one)=0.20 Answer
Answer
The possible exactly-one probability fills the interval from 0.20 through 0.90.
0.20≤Pr(exactly one)≤0.90(C)