Skip to content
View GOOD-123-CPU's full-sized avatar

Block or report GOOD-123-CPU

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
GOOD-123-CPU/README.md

David Zhu

数据科学与大数据技术 · Data Science and Big Data Technology

你好,我是 David Zhu。我关注可解释机器学习、统计与因果分析、知识检索和数据可视化,把分析方法落实为可复现的实验和可运行的应用。

I build reproducible data science projects and AI applications, with a focus on interpretable models, causal analysis, retrieval, and visualization.

精选项目 · 研究结果 · 更多探索 · 全部仓库

精选项目

项目 解决的问题 从这里开始
HanBayes 中文情感分类:贝叶斯基线、特征加权与可解释预测 模型对比结果 · 算法说明
VoxFrontier 直播效率分析:DEA、归因、Double ML 与反事实模拟 方法与假设 · 复现指南
MediRAG 医疗文档问答:多路检索、融合、重排序与引用溯源 检索链路 · 启动与评估边界
ScreenWeaver 配置驱动的数据大屏:Vue 3、TypeScript 与 ECharts 在线演示 · 完整配置示例
Retail Audit Agent 零售审计演示:资料解析、规则识别、AI 解释与报告流程 功能与本地体验
Research Figure Workbench 科研制图工作流:数据溯源、图表检查与可编辑导出 工作流与使用说明

研究结果

HanBayes · 中文情感分类

仓库已提交的模型对比结果:

模型 Accuracy Macro-F1 AUC
StandardNB 0.7793 0.7786 0.8505
FWNB 0.8022 0.8009 0.8770
DFWNB-v2 0.8048 0.8033 0.8822
SDFWNB 0.8073 0.8065 0.8867

SDFWNB 相比 StandardNB,Accuracy 提高 2.80 个百分点,Macro-F1 提高 0.0279。这些是仓库保存的实验结果,不是实时评测,也不代表在其他数据集上的表现。

HanBayes 模型对比图

原始结果 CSV · 冻结配置 · 复现入口

VoxFrontier · 效率分析与因果估计

从合成数据出发,连接 效率测度 → 贡献归因 → 因果估计 → 情景模拟,保留方法说明、结果表与运行记录。

VoxFrontier 合成数据分析总览

示例用于展示分析方法,不代表真实平台的实证结论;因果解释依赖识别假设,情景模拟也不等于经过验证的干预效果。

方法与假设 · 使用与复现

更多探索

交流

欢迎通过对应项目的 Issues 讨论方法、反馈问题或交流复现结果。项目的安装步骤、依赖和适用边界以各仓库文档为准。

Pinned Loading

  1. AlphaDebate-FinResearch AlphaDebate-FinResearch Public

    Financial research demo combining structured market data, multi-agent debate and report generation

    Python

  2. hanbayes hanbayes Public

    Interpretable Bayesian models for Chinese sentiment classification, with sparse features and reproducible evaluations

    Python

  3. medirag-open medirag-open Public

    Medical knowledge RAG demo with hybrid retrieval, reranking and source citations, built with Spring Boot, Vue and Milvus

    Java

  4. retail-audit-agent retail-audit-agent Public

    Retail audit demo combining document analysis, rule-based risk scoring, LLM explanations and review workflows

    TypeScript

  5. screenweaver screenweaver Public

    Configuration-driven data visualization engine: JSON layouts, reusable charts and live data sources with Vue 3 and ECharts

    TypeScript

  6. voxFrontier voxFrontier Public

    Reproducible efficiency and causal-analysis pipeline using synthetic livestream data: DEA, attribution, DML and counterfactual simulation

    Python