BIGDATA SAAS FOR INSURANCE BROKER | Unicsoft

Big Data SaaS for Insurance broker

BIGDATA SAAS FOR INSURANCE BROKER

Solution for a data-driven assessment

Situation

Insurance industry is a highly competitive market which requires its players to constantly improve their value proposition. However, simple short-term strategies such as reducing prices and rendering discounts may lead to huge losses in a long run. Therefore, in order to be successful in the market, insurance companies have to be more and more creative in building and perfecting robust and precise models for proper insurance rate calculations and risk assessments.

Being aware of successful cooperation between Unicsoft and one of Lead Automotive Dealer, an International Insurance Broker (further referred as IB) contacted Unicsoft with an idea to develop a solution for a data-driven assessment. Being a partner of several automotive dealers, IB has an access to data received from in-car gauges and in-car smart devices. Therefore, there are two sources of available data: historical and descriptive data about users and the above mentioned data coming from devices, which build the foundation for increasing accuracy of the final model through determining a driving style and User’s proneness to an accident. All the abovementioned components were evaluated by Unicsoft as being crucial in terms of their impact on the final model and insurance rate respectively.

 

Solution

Key points of solution for IB were defined as:

  • choice of proper solution for handling streams coming from two data sources (internal from IB and external from ADs);
  • data cleaning, noise and outliers reduction;
  • development a Scoring model for risk level and proneness assessment;
  • machine learning and model improvement within a specified period of time after its launch.

Based on the identified solution parameters and goals Unicsoft set a team of Advanced Analytics experts along with DevOps and Data Warehousing specialists, which completed initial data cleaning and predictive model within two months and a deployment-ready version of solution within seven months; further, a support team was set up and introduced to IB.

 

Result

The solution was successfully launched and had a commercial success with reaching the breakeven point within the first 18 months. The solution generated additional sales by offering a personal value proposition due to implementation of multi-feature risk assessment model and proper approach of rate evaluation for a particular User.

Having finished specified Model, IB underlined a very high level of Delivery process provided by Unicsoft, as well as a successful launch of the support team. Consequently, Unicsoft established a partnership with IB and was requested for development of yet another set of models outside Automotive domain.

TECHNOLOGY & TOOLS: PythonXGBoostMachine Learning

Platform: Web