Statistics · Model fitting · Easy
One regressor, one penalty
There are 100 observations of y and x. Both have a sample mean of zero, and x is scaled so that the sum of its squares is 100. The least-squares slope of y on x is 0.6. Instead you choose the slope b to minimise the sum of squared residuals plus 50 × b². What value of b does this give?
a fraction, an expression or a decimal to two significant figures · 5/8 · ln 2 · pi · inf
Answer and solution
Answer
2/5 ≈ 0.4
Solution
With one regressor a ridge penalty λ shrinks the least-squares slope by the factor Σx² / (Σx² + λ).
Setting the derivative to zero: Σxy = b (Σx² + 50). Σxy = 0.6 × 100 = 60, so b = 60 / 150 = 0.4.
Model fitting comes up in 6 firms' roundsTwo Sigma · Tower Research · Millennium · +3
These 8 rounds ask this kind of problem live, with follow-ups. Every round on every track is in Premium.