Original practice · fully worked
Original variant: weight two correlated sensors
Two sensors provide unbiased measurements A and B of the same temperature. Their errors have variances 9 and 4 and covariance 1. An unbiased combined estimate is wA+(1-w)B. Calculate the value of w that minimizes the estimate's error variance.
- A
0.000 - B
0.231 - C
0.273 - D
0.308 - E
0.500
Answer and calculation
Including covariance gives V(w)=9w²+4(1-w)²+2w(1-w)=11w²-6w+4. Its derivative vanishes at w=3/11=0.2727, so choice C.
Setup
Setup
The two weights sum to one, preserving unbiasedness. Apply the variance formula including the covariance term.
V(w)=9w2+4(1−w)2+2w(1−w)(1) Model
Model
Expand the expression into a quadratic.
V(w)=9w2+4−8w+4w2+2w−2w2 V(w)=11w2−6w+4 Compute
Compute
Solve the first-order condition and verify convexity.
V′(w)=22w−6=0 w=226=113=0.272727… V′′(w)=22>0 Answer
Answer
The higher-variance sensor receives weight approximately 0.273.
0.273(C)