System Architecture
Object-oriented design diagrams for the Insurance Analytics Platform. These diagrams illustrate the data model, ML pipeline, chat engine, component hierarchy, Python backend, and system architecture overview.
1. Data Model Layer
TypeScript interfaces defining the structure of all data flowing through the dashboard — metadata, quality metrics, statistical distributions, and correlation analysis.
2. ML Model & Text Analysis Layer
Machine learning pipeline outputs including multi-model comparison with 5-fold cross-validation, customer default rate prediction, topic classification, and text analysis.
3. Chat & Search Engine Layer
Client-side RAG engine with BM25 retrieval, SSE streaming for progressive LLM response display, and Kimi API integration.
4. React Component Hierarchy
Next.js app structure with 7 pages, shared chart/card components, and the chatbot interface with streaming markdown rendering.
5. Python Backend — Analysis Pipeline
Server-side analysis engine that processes 3 insurance datasets, trains multiple ML models with cross-validation, and builds the knowledge base.
6. System Architecture Overview
End-to-end data flow from CSV sources through Python analysis to the Next.js dashboard, with SSE streaming to the Kimi LLM API.