Sunshine Insurance Group Data Mining Platform Project
Sunshine Insurance Group Data Mining Platform Project
Project Background
In the insurance industry, as data accumulates over time, business data has exceeded the scale of hundreds of millions of records. Leveraging this valuable data for analysis and decision-making can generate significant added value. Amid the big data trend, Sunshine Insurance Group, one of China's seven major state-owned insurance groups, is actively harnessing new concepts and technologies to extract data value, enabling accurate and timely decision-making and continuously enhancing its competitiveness. The group's existing analysis system relied on a traditional database, whose performance could no longer meet the demands of querying, analyzing, and mining massive data. It urgently needed a new technical architecture to support complex analysis and in-depth mining of massive insurance business data.
Solution
To meet Sunshine Insurance Group's needs for querying, analyzing, and mining massive data, a data platform was built using the GBase 8a MPP Cluster, a large-scale distributed parallel database cluster system. The database handles the loading of massive data, the integration and common processing of fundamental data, and the construction of data marts for various themes, perfectly supporting the group's analytical applications and decision-making requirements.
Product Architecture Diagram
The project adopted a batch-generated ETL script tool to integrate data from various business application platforms, significantly improving development efficiency and ensuring data processing performance. This provides strong support for in-depth business analysis and meets the users' needs for insurance business system information integration and high-speed statistics.
The data mining platform supports multiple high-level applications for the group:
Analyzing historical policy information, customer information, transaction data, and financial data to improve the efficiency of new business development;
Leveraging insurance types, premium payment periods, insured occupations, annual incomes, and age groups to recommend optimal insurance products;
Performing big data combinations and mining to conduct in-depth analysis of target customer segments for insurance products.
Benefits
High-Speed Loading and Massive Storage: Achieves loading of tables with hundreds of millions of rows, along with a high compression ratio for ingestion to boost performance. It provides massive storage capacity, integrates data from multiple business departments, and supports dynamic online scaling as needed;
Ad-Hoc Queries with Sub-Second Response: Delivers high-speed ad-hoc and range queries on massive datasets, providing stable support for analysis systems;
High-Performance Analysis for Precision Operations: Significantly improves the computational and analytical performance of large table joins and multi-table join queries common in the insurance industry. Complex statistical analyses respond in seconds, enabling customers to achieve precision marketing and fine-grained operations.