机器学习开源金融模型Introduction / 项目简介 https://github.com/yhemeo613/

Introduction / 项目简介

This is an enterprise-grade automated financial model training and prediction system designed for quantitative trading and risk management. It integrates a modern React frontend with a robust FastAPI backend, leveraging advanced machine learning (Random Forest) and deep learning (LSTM) algorithms to analyze financial data, predict market trends, and assess risks.

这是一个企业级的自动化金融模型训练与预测系统，专为量化交易和风险管理设计。它集成了现代化的 React 前端和稳健的 FastAPI 后端，利用先进的机器学习（随机森林）和深度学习（LSTM）算法来分析金融数据、预测市场趋势并评估风险。

Key Features / 核心特性

Multi-Architecture Support / 多架构支持:
- LSTM (Long Short-Term Memory) : Deep learning model optimized for time-series forecasting with memory capabilities. Supports Dropout and L2 regularization.
- Random Forest: Ensemble learning method for robust regression and classification tasks.
- Ensemble Model: Weighted fusion of LSTM and Random Forest to reduce variance and improve prediction stability.
Advanced Training Pipeline / 高级训练流程:
- TimeSeriesSplit Cross-Validation: Strict time-series cross-validation to prevent look-ahead bias.
- Grid Search: Automated hyperparameter tuning for optimal model performance.
- Data Drift Detection: Real-time monitoring of data distribution changes using KS Test.
Comprehensive Risk Management / 全面风险管理:
- VaR (Value at Risk) : Calculation of potential loss at 95% confidence level.
- Volatility Analysis: Annualized volatility tracking.
- Drawdown Assessment: Historical maximum drawdown calculation.
Interactive Visualization / 交互式可视化:
- Real-time training progress monitoring.
- Interactive charts with confidence intervals (90% prediction bounds).
- Risk assessment dashboards.

🛠 Tech Stack / 技术栈

Frontend / 前端: React, Vite, Recharts, Lucide React, Axios
Backend / 后端: Python, FastAPI, Uvicorn, Pandas, Scikit-learn, PyTorch
Infrastructure / 基础设施: Docker, Docker Compose, Nginx

� Data Source & Configuration / 数据源与配置

Data Source / 数据来源

The system fetches real-time and historical financial data from Binance (world's leading cryptocurrency exchange). The data pipeline handles:

Market data acquisition (Kline/Candlestick data)
Technical indicator calculation (MA, RSI, MACD, Bollinger Bands)
Data cleaning and normalization

本系统的数据来源于 Binance (币安) 交易所。数据管道负责处理：

市场行情数据获取 (K线数据)
技术指标计算 (MA, RSI, MACD, 布林带等)
数据清洗与归一化

Environment Configuration / 环境配置

To access the data source, you need to configure your API keys in a .env file at the project root. 为了访问数据源，您需要在项目根目录创建一个 .env 文件并配置 API 密钥。

Example .env file / .env 文件示例:

BINANCE_API_KEY=your_api_key_here
BINANCE_SECRET_KEY=your_secret_key_here

Note: Ensure your API keys have read-only permissions for security. 注意：请确保您的 API 密钥仅拥有只读权限以保证安全。

�🚀 Getting Started / 快速开始

Prerequisites / 前置要求

Docker & Docker Compose
Node.js 18+ (for local development)
Python 3.10+ (for local development)

Deployment with Docker / 使用 Docker 部署

Clone the repository / 克隆仓库:

git clone <repository-url>
cd <project-directory>

Build and Run / 构建并运行:
```
docker-compose up --build -d
```
Access the Application / 访问应用:
- Frontend: http://localhost:80
- Backend API Docs: http://localhost:8000/docs

Local Development / 本地开发

Backend / 后端

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run server
uvicorn system_integration.api.main:app --reload

Frontend / 前端

cd frontend

# Install dependencies
npm install

# Run dev server
npm run dev

📊 Model Configuration / 模型配置

Parameter	Description	Default	Applicable Models
Epochs	Number of training iterations	10	LSTM, Ensemble
Learning Rate (LR)	Step size for optimizer	0.001	LSTM, Ensemble
Dropout	Regularization rate	0.2	LSTM, Ensemble
N Estimators	Number of trees in the forest	100	Random Forest, Ensemble
Days	Prediction horizon	30	All

机器学习 开源金融模型