Insurance company scoring
About project
Bisiness effect
The expected tariffs for clients and the forecast of unprofitable cases at the beginning of work are calculated. The company became more productive in the insurance market: sales increased by 10%.
Solution
Building models to predict each case
A task
Models need to be built to predict customer losses, the size of the loss, and the likelihood of buying and renewing a policy.
Stages of development
Тechnology
Development of a pilot model
2.
Productivity increase
3.
Model Implementation
4.
Data analysis
1.
Model monitoring
5.
A set of libraries for building GLM H2O models
The generalized linear model (GLiM or GLM) is an advanced statistical modeling technique. Used in all applications of applied statistics.
Libraries for building, boosting models
Lightgbm, CatBoost, XGBoost
Insolver.io
An auto ML library to build and implement models in a fast way
Project team
Engineer
Alexander H.
Аnalyst
Alexander S.
Team lead
Frank Sh.
Areas of use
This technology is needed by insurance companies. In order to calculate the losses, the amount of losses, the probability of buying a policy, the probability of extending the contract.
These models will help to calculate tariffs for the client and be more competitive in the insurance market.
These models will help to calculate tariffs for the client and be more competitive in the insurance market.
