Ranked in top 10% of registered competitors in McKinsey Analytics Online Hackathon - 20th - 22nd July 2018.
From labelled data about an insurance policy, predict the probability of renewal of a policy.
Find incentives for each policy to generate maximum revenue given following relationships:
Incentive vs Efforts: Effort = 10 * ( 1 - e^(-Incentive/400) )
Efforts vs % improvement in renewal probability: % improvement in renewal probability = 20 * ( 1 - e^(-Effort/5) )
Total Revenue = Sum( (p_benchmark * delta_p) * premium - Incentive ) where,
p_benchmark = benchmark probability defined by the client. (The closer your values of renewal probability are to the true values, the closer it is to the benchmark probability),
delta_p = (1 + % improvement in probability of renewal),
premium = premium on policy,
Incentive = Incentive given to agent to increase renewal probability