Machine Learning Platform for Insurance Cases
The Machine Learning Platform consist of a Kerberized Cloudera Hadoop Cluster with Hive, Spark and Cloudera Workbench. Data Reply continues expanding the capabilities of the platform and provides 3rd level DevOps support.
At the same time the customers Data Science teams and Data Reply Data Scientists are jointly developing and testing several Machine Learning use cases, providing new insights for the different business units of the insurance company.
Use cases include:
- Complexity Reduction in Risk Identification Processes based on intelligent usage of customer data
- Short Term Incident Prediction - Forecasting damages for home insurance based on the weather forecast
- Pricing Optimization through supervised machine learning allowing to increase cost efficiency and higher accuracy in risk assessment
- Improved existing fraud detection models via machine learning and intelligent suggestions of potential fraud cases in future models with high success rates
- Optimized Input Management to allow for automatic company specific classification of documents reducing classification errors and manual effort
Positive business outcomes
Our use cases demonstrate fields of improved customer excellence, cost-efficiency and sustained business model development via reduced complexity, service availability and alignment of processes. The Machine Learning Platform as set up by Data Reply enables the customer to improve its services, work more efficient and get insights faster. Additionally, the customer can provide Data Science Support and Resources to others, opening new fields of business and revenue.
Metrics and proof points
- Fraud case clustering allows for successful identification of clusters with 1-in-3 probability of fraud cases
- Five times higher success rate in identifying fraud cases correctly compared to previous methodologies
- Meeting SLAs in claims-hotlines despite mass-events compared to previous breakdown-events
- Optimized claims management and liquidity due to improved prediction quality
- Availability of highly skilled data scientist and engineers
- Investing in new talent, resources and technologies in our outstanding data incubator program
- Deep understanding of industry specific requirements and challenges
- Elevating business impact and sustainably implementation through E2E delivery and long-term partnerships
Drive Customer Insights
About Data Reply
Data Reply is a Consultancy focused heavily on Distributed Open Source & Cloud Technologies and Data Science. Data Reply is not focused on a specific industry rather focused on use-cases and customized solutions which are highly automated and scalable.
Cloudera, Kafka, Hadoop, Spark (pySpark, Scala, Sparklyr), Hive, Kerberos, Kubernetes, Cloudera Workbench, Ansible