I recently got an abstract case about solving the problem of increased crime rate against women in a major metropolitan city. The case had a market sizing part to estimate the number of crimes against women. I went with a simple approach of doing a population estimation of women, segmenting them based on age, estimating the potential of each segment facing crime (Women between 20 - 40 are more prone to harassment), then took an estimate of how many of these women actually face these crimes. But I felt like it was too simple and could be more sophisticated. Would love to hear different appraoches, thanks.ย
Estimate crime rate against women in a major city
As an introduction, let me say that the best possible approach for market sizing is the one that maximizes Simplicity * correctness i.e. gives you a feasible range for the answer and directs you towards the correct action / solution for the client using less data, assumptions...
That said, to be more precise you could further segment the crime rate on the basis of the size of the cities, e.g. metropolitan cities vs. medium cities, vs. small villages and change accordingly your "conversion rate" for the women that actually face these crimes (e.g. metropolitan cities 50% more vs. small village)
You used a "demand side" approach to the market sizing, which to me, seems pretty good choice, but you could have used a supply side approach
E.g. (high level),ย you could estimate the number of crimes in each type of cities based on drivers. You could start from the population of each type of cities, segment by age and assign the probability to be a criminal (crimes / population) and then assign a % of this crime to be a crime against women.
You can further sopistichate this approach, too