Estimate Event Rates from Unequal-Probability Samples
WaymoData Scientist · Technical Screen
Last updated
Establish event rates from unequal sampling with inverse-probability weighting, exposure-aware denominators, and efficient stratified allocation.
Estimate Event Rates from Unequal-Probability Samples
You have a population table of units and their known probabilities of being selected for inspection. A second table records which units were inspected and whether an event occurred. Explain how to estimate the population event rate.
A related logging system retains all nighttime observations but only a known random fraction of daytime observations. Explain how to estimate the overall weekly event rate and how you would choose sampling fractions across strata if you controlled the logging budget. Clarify the event-rate denominator rather than assuming that a unit rate and a per-mile rate are identical.
Hints
Hint 1
Restore what was less likely to be seen: An observed unit selected with probability one tenth represents a different amount of population information from a census unit.
Follow-up questions
- Why is the raw event fraction among inspected units generally biased for the population fraction?
- What cannot be recovered if a relevant group has zero chance of inclusion?
Solution walkthrough is for members
Get the full walkthrough with reference code and complexity analysis, plus Kitty Coach at 300 messages a day. Hints, the code runner and community answers stay free.