[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-feature-engine--feature_engine":3,"tool-feature-engine--feature_engine":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":79,"owner_twitter":79,"owner_website":79,"owner_url":80,"languages":81,"stars":90,"forks":91,"last_commit_at":92,"license":93,"difficulty_score":94,"env_os":95,"env_gpu":96,"env_ram":96,"env_deps":97,"category_tags":104,"github_topics":105,"view_count":112,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":113,"updated_at":114,"faqs":115,"releases":146},131,"feature-engine\u002Ffeature_engine","feature_engine","Feature engineering and selection open-source Python library compatible with sklearn.","feature_engine 是一个开源的 Python 库，专注于机器学习中的特征工程与特征选择。它提供了一系列易于使用的转换器，帮助用户对原始数据进行清洗、编码、变换和筛选，从而构建更适合模型训练的输入特征。这些操作包括处理缺失值、变量编码（如独热编码、目标编码）、变量变换（如对数、幂变换）、离群值处理以及基于统计或模型的特征选择等。\n\nfeature_engine 的设计完全兼容 scikit-learn，所有转换器都遵循 fit() 和 transform() 接口规范，可无缝集成到标准的机器学习流水线中，解决了数据预处理流程难以复用和部署的问题。它特别适合从事数据科学和机器学习的开发者、研究人员及数据工程师使用，尤其适用于需要构建稳健、可维护特征工程流程的项目。\n\n其技术亮点在于将业界常用的特征工程方法封装成标准化组件，既保证了灵活性，又提升了代码的可读性与可复现性。无论是初学者还是经验丰富的从业者，都能借助 feature_engine 更高效地完成从原始数据到模型就绪特征的转化。","# Feature-engine\n\n[![feature-engine logo](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ffeature-engine_feature_engine_readme_24cec6e8fc9e.png)](http:\u002F\u002Ffeature-engine.readthedocs.io)\n\n| | |\n| --- | --- |\n| **Open Source** | [![GitHub](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Ffeature-engine\u002Ffeature_engine)](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fblob\u002Fmaster\u002FLICENSE.md)  [![Sponsorship](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPowered%20By-TrainInData-orange.svg)](https:\u002F\u002Fwww.trainindata.com\u002F) |\n| **Tutorials** | [![!youtube](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?logo=youtube&label=YouTube&message=tutorials&color=red)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=nisuJQTyDSA&list=PL_7uaHXkQmKVXh06fCWRxdQRZsl3wva5k) |\n| **Code** | [![PyPI - Python Version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Ffeature_engine?logo=Python)](https:\u002F\u002Fpypi.org\u002Fproject\u002Ffeature-engine\u002F) [![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Ffeature_engine?logo=PyPI)](https:\u002F\u002Fpypi.org\u002Fproject\u002Ffeature-engine) [![Conda](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fv\u002Fconda-forge\u002Ffeature_engine?logo=Anaconda)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Ffeature_engine) |\n| **Downloads** | [![Monthly Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Ffeature-engine)](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Ffeature-engine)  [![Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ffeature-engine_feature_engine_readme_bb019999e6b4.png))](https:\u002F\u002Fpepy.tech\u002Fproject\u002Ffeature-engine)|\n| **Meta** | [![GitHub contributors](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Ffeature-engine\u002Ffeature_engine?logo=GitHub)](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fgraphs\u002Fcontributors) [![first-timers-only](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ffirst--timers--only-friendly-blue.svg?style=flat)](https:\u002F\u002Fwww.firsttimersonly.com\u002F) |\n| **Documentation** | [![Read the Docs](https:\u002F\u002Fimg.shields.io\u002Freadthedocs\u002Ffeature_engine?logo=readthedocs)](https:\u002F\u002Ffeature-engine.readthedocs.io\u002Fen\u002Flatest\u002Findex.html) |\n| **Citation** | [![DOI](https:\u002F\u002Fzenodo.org\u002Fbadge\u002F163630824.svg)](https:\u002F\u002Fzenodo.org\u002Fbadge\u002Flatestdoi\u002F163630824) [![JOSS](https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.03642\u002Fstatus.svg)](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.03642) |\n| **Testing** | [![CircleCI](https:\u002F\u002Fimg.shields.io\u002Fcircleci\u002Fbuild\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine\u002Fmain?logo=CircleCI)](https:\u002F\u002Fapp.circleci.com\u002Fpipelines\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine) [![Codecov](https:\u002F\u002Fimg.shields.io\u002Fcodecov\u002Fc\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine?logo=CodeCov&token=ZBKKSN6ERL)](https:\u002F\u002Fcodecov.io\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine) [![Code style: black](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-000000.svg)](https:\u002F\u002Fgithub.com\u002Fpsf\u002Fblack) |\n\u003Cdiv align=\"center\">\n\n\n\u003C\u002Fdiv>\n\nFeature-engine is a Python library with multiple transformers to engineer and select features for use in machine learning models. \nFeature-engine's transformers follow Scikit-learn's functionality with fit() and transform() methods to learn the \ntransforming parameters from the data and then transform it.\n\n\n## Feature-engine features in the following resources\n\n* [Feature Engineering for Machine Learning, Online Course](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-engineering-for-machine-learning)\n\n* [Feature Selection for Machine Learning, Online Course](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-selection-for-machine-learning)\n\n* [Feature Engineering for Time Series Forecasting, Online Course](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-engineering-for-forecasting)\n\n* [Forecasting with Machine Learning, Online Course](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Fforecasting-with-machine-learning)\n\n* [Python Feature Engineering Cookbook](https:\u002F\u002Fwww.packtpub.com\u002Fen-us\u002Fproduct\u002Fpython-feature-engineering-cookbook-9781835883587)\n\n* [Feature Selection in Machine Learning Book](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-selection-in-machine-learning-book)\n\n\n## Blogs about Feature-engine\n\n* [Feature-engine: A new open-source Python package for feature engineering](https:\u002F\u002Ftrainindata.medium.com\u002Ffeature-engine-a-new-open-source-python-package-for-feature-engineering-29a0ab88ea7c)\n\n* [Practical Code Implementations of Feature Engineering for Machine Learning with Python](https:\u002F\u002Ftowardsdatascience.com\u002Fpractical-code-implementations-of-feature-engineering-for-machine-learning-with-python-f13b953d4bcd)\n\n\n## Documentation\n\n* [Documentation](https:\u002F\u002Ffeature-engine.trainindata.com)\n\n\n## Pst! How did you find us?\n\nWe want to share Feature-engine with more people. It'd help us loads if you tell us\nhow you discovered us.\n\nThen we'd know what we are doing right and which channels to use to share the love.\n\nPlease share your story by answering 1 quick question\n[at this link](https:\u002F\u002Fdocs.google.com\u002Fforms\u002Fd\u002Fe\u002F1FAIpQLSfxvgnJvuvPf2XgosakhXo5VNQafqRrjNXkoW5qDWqnuxZNSQ\u002Fviewform?usp=sf_link)\n. 😃\n\n## Current Feature-engine's transformers include functionality for:\n\n* Missing Data Imputation\n* Categorical Encoding\n* Discretisation\n* Outlier Capping or Removal\n* Variable Transformation\n* Variable Creation\n* Variable Selection\n* Datetime Features\n* Text Features\n* Time Series\n* Preprocessing\n* Scaling\n* Scikit-learn Wrappers\n\n### Imputation Methods\n* MeanMedianImputer\n* ArbitraryNumberImputer\n* RandomSampleImputer\n* EndTailImputer\n* CategoricalImputer\n* AddMissingIndicator\n* DropMissingData\n\n### Encoding Methods\n* OneHotEncoder\n* OrdinalEncoder\n* CountFrequencyEncoder\n* MeanEncoder\n* WoEEncoder\n* RareLabelEncoder\n* DecisionTreeEncoder\n* StringSimilarityEncoder\n\n### Discretisation methods\n* EqualFrequencyDiscretiser\n* EqualWidthDiscretiser\n* GeometricWidthDiscretiser\n* DecisionTreeDiscretiser\n* ArbitraryDiscreriser\n\n### Outlier Handling methods\n* Winsorizer\n* ArbitraryOutlierCapper\n* OutlierTrimmer\n\n### Variable Transformation methods\n* LogTransformer\n* LogCpTransformer\n* ReciprocalTransformer\n* ArcsinTransformer\n* PowerTransformer\n* BoxCoxTransformer\n* YeoJohnsonTransformer\n* ArcSinhTransformer\n\n### Variable Scaling methods\n* MeanNormalizationScaler\n\n### Variable Creation:\n * MathFeatures\n * RelativeFeatures\n * CyclicalFeatures\n * DecisionTreeFeatures\n * GeoDistanceFeatures\n\n### Feature Selection:\n * DropFeatures\n * DropConstantFeatures\n * DropDuplicateFeatures\n * DropCorrelatedFeatures\n * SmartCorrelationSelection\n * ShuffleFeaturesSelector\n * SelectBySingleFeaturePerformance\n * SelectByTargetMeanPerformance\n * RecursiveFeatureElimination\n * RecursiveFeatureAddition\n * DropHighPSIFeatures\n * SelectByInformationValue\n * ProbeFeatureSelection\n * MRMR\n\n### Datetime\n * DatetimeFeatures\n * DatetimeSubtraction\n * DatetimeOrdinal\n\n### Text Features\n * TextFeatures\n \n### Time Series\n * LagFeatures\n * WindowFeatures\n * ExpandingWindowFeatures\n \n### Pipelines\n* Pipeline\n* make_pipeline\n\n### Preprocessing\n * MatchCategories\n * MatchVariables\n \n### Wrappers:\n * SklearnTransformerWrapper\n\n## Installation\n\nFrom PyPI using pip:\n\n```\npip install feature_engine\n```\n\nFrom Anaconda:\n\n```\nconda install -c conda-forge feature_engine\n```\n\nOr simply clone it:\n\n```\ngit clone https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine.git\n```\n\n## Example Usage\n\n```python\n>>> import pandas as pd\n>>> from feature_engine.encoding import RareLabelEncoder\n\n>>> data = {'var_A': ['A'] * 10 + ['B'] * 10 + ['C'] * 2 + ['D'] * 1}\n>>> data = pd.DataFrame(data)\n>>> data['var_A'].value_counts()\n```\n\n```\nOut[1]:\nA    10\nB    10\nC     2\nD     1\nName: var_A, dtype: int64\n```\n    \n```python \n>>> rare_encoder = RareLabelEncoder(tol=0.10, n_categories=3)\n>>> data_encoded = rare_encoder.fit_transform(data)\n>>> data_encoded['var_A'].value_counts()\n```\n\n```\nOut[2]:\nA       10\nB       10\nRare     3\nName: var_A, dtype: int64\n```\n\nFind more examples in our [Jupyter Notebook Gallery](https:\u002F\u002Fnbviewer.org\u002Fgithub\u002Ffeature-engine\u002Ffeature-engine-examples\u002Ftree\u002Fmain\u002F) \nor in the [documentation](https:\u002F\u002Ffeature-engine.trainindata.com).\n\n## Contribute\n\nDetails about how to contribute can be found in the [Contribute Page](https:\u002F\u002Ffeature-engine.trainindata.com\u002Fen\u002Flatest\u002Fcontribute\u002Findex.html)\n\nBriefly:\n\n- Fork the repo\n- Clone your fork into your local computer:\n```\ngit clone https:\u002F\u002Fgithub.com\u002F\u003CYOURUSERNAME>\u002Ffeature_engine.git\n```\n- navigate into the repo folder\n```\ncd feature_engine\n```\n- Install Feature-engine as a developer: \n```\npip install -e .\n```\n- Optional: Create and activate a virtual environment with any tool of choice\n- Install Feature-engine developer dependencies: \n```\npip install -e \".[tests]\"\n``` \n- Create a feature branch with a meaningful name for your feature: \n```\ngit checkout -b myfeaturebranch\n```\n- Develop your feature, tests and documentation\n- Make sure the tests pass\n- Make a PR\n\nThank you!!\n\n\n### Documentation\n\nFeature-engine documentation is built using [Sphinx](https:\u002F\u002Fwww.sphinx-doc.org) and is hosted on [Read the Docs](https:\u002F\u002Freadthedocs.org\u002F).\n\nTo build the documentation make sure you have the dependencies installed: from the root directory: \n```\npip install -r docs\u002Frequirements.txt\n```\n\nNow you can build the docs using: \n```\nsphinx-build -b html docs build\n```\n\n\n## License\n\nThe content of this repository is licensed under a [BSD 3-Clause license](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fblob\u002Fmain\u002FLICENSE.md).\n\n\n## Sponsor\n\nFeature-engine is made possible with the support of [Train in Data](https:\u002F\u002Fwww.trainindata.com\u002F).\n","# Feature-engine\n\n[![feature-engine logo](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ffeature-engine_feature_engine_readme_24cec6e8fc9e.png)](http:\u002F\u002Ffeature-engine.readthedocs.io)\n\n| | |\n| --- | --- |\n| **开源** | [![GitHub](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Ffeature-engine\u002Ffeature_engine)](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fblob\u002Fmaster\u002FLICENSE.md)  [![Sponsorship](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPowered%20By-TrainInData-orange.svg)](https:\u002F\u002Fwww.trainindata.com\u002F) |\n| **教程** | [![!youtube](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?logo=youtube&label=YouTube&message=tutorials&color=red)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=nisuJQTyDSA&list=PL_7uaHXkQmKVXh06fCWRxdQRZsl3wva5k) |\n| **代码** | [![PyPI - Python Version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Ffeature_engine?logo=Python)](https:\u002F\u002Fpypi.org\u002Fproject\u002Ffeature-engine\u002F) [![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Ffeature_engine?logo=PyPI)](https:\u002F\u002Fpypi.org\u002Fproject\u002Ffeature-engine) [![Conda](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fv\u002Fconda-forge\u002Ffeature_engine?logo=Anaconda)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Ffeature_engine) |\n| **下载量** | [![Monthly Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Ffeature-engine)](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Ffeature-engine)  [![Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ffeature-engine_feature_engine_readme_bb019999e6b4.png))](https:\u002F\u002Fpepy.tech\u002Fproject\u002Ffeature-engine)|\n| **元信息** | [![GitHub contributors](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Ffeature-engine\u002Ffeature_engine?logo=GitHub)](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fgraphs\u002Fcontributors) [![first-timers-only](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ffirst--timers--only-friendly-blue.svg?style=flat)](https:\u002F\u002Fwww.firsttimersonly.com\u002F) |\n| **文档** | [![Read the Docs](https:\u002F\u002Fimg.shields.io\u002Freadthedocs\u002Ffeature_engine?logo=readthedocs)](https:\u002F\u002Ffeature-engine.readthedocs.io\u002Fen\u002Flatest\u002Findex.html) |\n| **引用** | [![DOI](https:\u002F\u002Fzenodo.org\u002Fbadge\u002F163630824.svg)](https:\u002F\u002Fzenodo.org\u002Fbadge\u002Flatestdoi\u002F163630824) [![JOSS](https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.03642\u002Fstatus.svg)](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.03642) |\n| **测试** | [![CircleCI](https:\u002F\u002Fimg.shields.io\u002Fcircleci\u002Fbuild\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine\u002Fmain?logo=CircleCI)](https:\u002F\u002Fapp.circleci.com\u002Fpipelines\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine) [![Codecov](https:\u002F\u002Fimg.shields.io\u002Fcodecov\u002Fc\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine?logo=CodeCov&token=ZBKKSN6ERL)](https:\u002F\u002Fcodecov.io\u002Fgithub\u002Ffeature-engine\u002Ffeature_engine) [![Code style: black](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-000000.svg)](https:\u002F\u002Fgithub.com\u002Fpsf\u002Fblack) |\n\u003Cdiv align=\"center\">\n\n\n\u003C\u002Fdiv>\n\nFeature-engine 是一个 Python 库，提供了多种用于机器学习模型的特征工程（Feature Engineering）和特征选择（Feature Selection）转换器（transformer）。  \nFeature-engine 的转换器遵循 Scikit-learn 的功能设计，提供 fit() 和 transform() 方法，先从数据中学习转换参数，然后对数据进行转换。\n\n## Feature-engine 出现在以下资源中\n\n* [《机器学习中的特征工程》在线课程](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-engineering-for-machine-learning)\n\n* [《机器学习中的特征选择》在线课程](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-selection-for-machine-learning)\n\n* [《时间序列预测中的特征工程》在线课程](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-engineering-for-forecasting)\n\n* [《使用机器学习进行预测》在线课程](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Fforecasting-with-machine-learning)\n\n* [《Python 特征工程实战手册》](https:\u002F\u002Fwww.packtpub.com\u002Fen-us\u002Fproduct\u002Fpython-feature-engineering-cookbook-9781835883587)\n\n* [《机器学习中的特征选择》书籍](https:\u002F\u002Fwww.trainindata.com\u002Fp\u002Ffeature-selection-in-machine-learning-book)\n\n\n## 关于 Feature-engine 的博客文章\n\n* [Feature-engine：一个用于特征工程的新开源 Python 包](https:\u002F\u002Ftrainindata.medium.com\u002Ffeature-engine-a-new-open-source-python-package-for-feature-engineering-29a0ab88ea7c)\n\n* [使用 Python 实现机器学习特征工程的实用代码示例](https:\u002F\u002Ftowardsdatascience.com\u002Fpractical-code-implementations-of-feature-engineering-for-machine-learning-with-python-f13b953d4bcd)\n\n\n## 文档\n\n* [官方文档](https:\u002F\u002Ffeature-engine.trainindata.com)\n\n\n## 嘿！你是怎么找到我们的？\n\n我们希望让更多人了解 Feature-engine。如果你能告诉我们你是如何发现我们的，将对我们帮助巨大。\n\n这样我们就能知道哪些做法是有效的，并选择合适的渠道继续传播。\n\n请通过点击[此链接](https:\u002F\u002Fdocs.google.com\u002Fforms\u002Fd\u002Fe\u002F1FAIpQLSfxvgnJvuvPf2XgosakhXo5VNQafqRrjNXkoW5qDWqnuxZNSQ\u002Fviewform?usp=sf_link)回答一个简单的问题来分享你的故事吧！😃\n\n## 当前 Feature-engine 的转换器包含以下功能：\n\n* 缺失值填补（Missing Data Imputation）\n* 类别编码（Categorical Encoding）\n* 离散化（Discretisation）\n* 异常值处理（Outlier Capping or Removal）\n* 变量变换（Variable Transformation）\n* 变量创建（Variable Creation）\n* 变量选择（Variable Selection）\n* 日期时间特征（Datetime Features）\n* 文本特征（Text Features）\n* 时间序列（Time Series）\n* 预处理（Preprocessing）\n* 缩放（Scaling）\n* Scikit-learn 封装器（Scikit-learn Wrappers）\n\n### 填补方法（Imputation Methods）\n* MeanMedianImputer\n* ArbitraryNumberImputer\n* RandomSampleImputer\n* EndTailImputer\n* CategoricalImputer\n* AddMissingIndicator\n* DropMissingData\n\n### 编码方法（Encoding Methods）\n* OneHotEncoder\n* OrdinalEncoder\n* CountFrequencyEncoder\n* MeanEncoder\n* WoEEncoder\n* RareLabelEncoder\n* DecisionTreeEncoder\n* StringSimilarityEncoder\n\n### 离散化方法（Discretisation methods）\n* EqualFrequencyDiscretiser\n* EqualWidthDiscretiser\n* GeometricWidthDiscretiser\n* DecisionTreeDiscretiser\n* ArbitraryDiscreriser\n\n### 异常值处理方法（Outlier Handling methods）\n* Winsorizer\n* ArbitraryOutlierCapper\n* OutlierTrimmer\n\n### 变量变换方法（Variable Transformation methods）\n* LogTransformer\n* LogCpTransformer\n* ReciprocalTransformer\n* ArcsinTransformer\n* PowerTransformer\n* BoxCoxTransformer\n* YeoJohnsonTransformer\n* ArcSinhTransformer\n\n### 变量缩放方法（Variable Scaling methods）\n* MeanNormalizationScaler\n\n### 变量创建（Variable Creation）:\n * MathFeatures\n * RelativeFeatures\n * CyclicalFeatures\n * DecisionTreeFeatures\n * GeoDistanceFeatures\n\n### 特征选择（Feature Selection）:\n * DropFeatures\n * DropConstantFeatures\n * DropDuplicateFeatures\n * DropCorrelatedFeatures\n * SmartCorrelationSelection\n * ShuffleFeaturesSelector\n * SelectBySingleFeaturePerformance\n * SelectByTargetMeanPerformance\n * RecursiveFeatureElimination\n * RecursiveFeatureAddition\n * DropHighPSIFeatures\n * SelectByInformationValue\n * ProbeFeatureSelection\n * MRMR\n\n### 日期时间（Datetime）\n * DatetimeFeatures\n * DatetimeSubtraction\n * DatetimeOrdinal\n\n### 文本特征（Text Features）\n * TextFeatures\n \n### 时间序列（Time Series）\n * LagFeatures\n * WindowFeatures\n * ExpandingWindowFeatures\n \n### 流水线（Pipelines）\n* Pipeline\n* make_pipeline\n\n### 预处理（Preprocessing）\n * MatchCategories\n * MatchVariables\n \n### 封装器（Wrappers）:\n * SklearnTransformerWrapper\n\n## 安装\n\n通过 PyPI 使用 pip 安装：\n\n```\npip install feature_engine\n```\n\n通过 Anaconda 安装：\n\n```\nconda install -c conda-forge feature_engine\n```\n\n或者直接克隆仓库：\n\n```\ngit clone https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine.git\n```\n\n## 示例用法\n\n```python\n>>> import pandas as pd\n>>> from feature_engine.encoding import RareLabelEncoder\n\n>>> data = {'var_A': ['A'] * 10 + ['B'] * 10 + ['C'] * 2 + ['D'] * 1}\n>>> data = pd.DataFrame(data)\n>>> data['var_A'].value_counts()\n```\n\n```\nOut[1]:\nA    10\nB    10\nC     2\nD     1\nName: var_A, dtype: int64\n```\n    \n```python \n>>> rare_encoder = RareLabelEncoder(tol=0.10, n_categories=3)\n>>> data_encoded = rare_encoder.fit_transform(data)\n>>> data_encoded['var_A'].value_counts()\n```\n\n```\nOut[2]:\nA       10\nB       10\nRare     3\nName: var_A, dtype: int64\n```\n\n更多示例请参见我们的 [Jupyter Notebook Gallery](https:\u002F\u002Fnbviewer.org\u002Fgithub\u002Ffeature-engine\u002Ffeature-engine-examples\u002Ftree\u002Fmain\u002F)  \n或 [文档](https:\u002F\u002Ffeature-engine.trainindata.com)。\n\n## 贡献\n\n有关如何贡献的详细信息，请参阅 [贡献页面](https:\u002F\u002Ffeature-engine.trainindata.com\u002Fen\u002Flatest\u002Fcontribute\u002Findex.html)。\n\n简要步骤如下：\n\n- Fork 本仓库\n- 将你的 fork 克隆到本地计算机：\n```\ngit clone https:\u002F\u002Fgithub.com\u002F\u003CYOURUSERNAME>\u002Ffeature_engine.git\n```\n- 进入仓库目录：\n```\ncd feature_engine\n```\n- 以开发者模式安装 Feature-engine：\n```\npip install -e .\n```\n- （可选）使用你喜欢的工具创建并激活虚拟环境\n- 安装 Feature-engine 的开发依赖项：\n```\npip install -e \".[tests]\"\n``` \n- 创建一个具有有意义名称的功能分支：\n```\ngit checkout -b myfeaturebranch\n```\n- 开发你的功能、测试和文档\n- 确保所有测试通过\n- 提交 Pull Request (PR)\n\n非常感谢！！\n\n### 文档\n\nFeature-engine 的文档使用 [Sphinx](https:\u002F\u002Fwww.sphinx-doc.org) 构建，并托管在 [Read the Docs](https:\u002F\u002Freadthedocs.org\u002F) 上。\n\n要构建文档，请确保已安装相关依赖项：从项目根目录运行以下命令：\n```\npip install -r docs\u002Frequirements.txt\n```\n\n然后即可使用以下命令构建文档：\n```\nsphinx-build -b html docs build\n```\n\n\n## 许可证\n\n本仓库的内容采用 [BSD 3-Clause 许可证](https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fblob\u002Fmain\u002FLICENSE.md) 授权。\n\n\n## 赞助方\n\nFeature-engine 的开发得到了 [Train in Data](https:\u002F\u002Fwww.trainindata.com\u002F) 的支持。","# Feature-engine 快速上手指南\n\n## 环境准备\n\n- **Python 版本**：3.7 或更高版本  \n- **依赖库**：Feature-engine 基于 scikit-learn 构建，安装时会自动处理依赖  \n- **推荐环境**：建议使用虚拟环境（如 `venv` 或 `conda`）隔离项目依赖\n\n> 💡 国内用户可配置 pip 镜像源加速下载，例如：\n> ```bash\n> pip install -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple feature_engine\n> ```\n\n## 安装步骤\n\n### 使用 pip 安装（推荐）\n```bash\npip install feature_engine\n```\n\n### 使用 conda 安装\n```bash\nconda install -c conda-forge feature_engine\n```\n\n### 从源码安装（开发用途）\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine.git\ncd feature_engine\npip install -e .\n```\n\n## 基本使用\n\n以下示例展示如何使用 `RareLabelEncoder` 将低频类别合并为 \"Rare\"：\n\n```python\n>>> import pandas as pd\n>>> from feature_engine.encoding import RareLabelEncoder\n\n>>> data = {'var_A': ['A'] * 10 + ['B'] * 10 + ['C'] * 2 + ['D'] * 1}\n>>> data = pd.DataFrame(data)\n>>> data['var_A'].value_counts()\nA    10\nB    10\nC     2\nD     1\nName: var_A, dtype: int64\n\n>>> rare_encoder = RareLabelEncoder(tol=0.10, n_categories=3)\n>>> data_encoded = rare_encoder.fit_transform(data)\n>>> data_encoded['var_A'].value_counts()\nA       10\nB       10\nRare     3\nName: var_A, dtype: int64\n```\n\n> 更多示例请参考官方 [Jupyter Notebook 示例库](https:\u002F\u002Fnbviewer.org\u002Fgithub\u002Ffeature-engine\u002Ffeature-engine-examples\u002Ftree\u002Fmain\u002F) 或 [中文文档](https:\u002F\u002Ffeature-engine.trainindata.com)。","某电商公司的数据科学团队正在构建用户复购预测模型，需对包含大量缺失值、类别变量和异常值的原始交易日志进行特征工程。\n\n### 没有 feature_engine 时\n- 需手动编写代码处理不同类型的缺失值（如数值型用中位数、分类型用“Missing”填充），逻辑分散且易出错  \n- 对高基数类别变量（如商品ID）进行频次编码或目标编码时，需自行实现并确保训练\u002F测试集一致性  \n- 异常值处理依赖临时脚本，难以复用，且无法与 sklearn Pipeline 无缝集成  \n- 特征选择（如剔除低方差或高相关性特征）需组合多个库，代码冗长且维护困难  \n\n### 使用 feature_engine 后\n- 利用 `MeanMedianImputer` 和 `CategoricalImputer` 等专用类，一行代码即可按变量类型自动填充缺失值  \n- 通过 `FrequencyEncoder` 或 `TargetEncoder` 直接完成高基数类别变量转换，并天然支持 fit\u002Ftransform 分离，避免数据泄露  \n- 使用 `Winsorizer` 或 `ArbitraryOutlierCapper` 统一处理异常值，参数可学习、过程可复现  \n- 借助 `DropConstantFeatures`、`SmartCorrelatedSelection` 等选择器，轻松嵌入 sklearn Pipeline 实现端到端流程  \n\nfeature_engine 将繁琐、易错的手工特征工程转化为标准化、可复用且与 sklearn 兼容的组件，显著提升建模效率与代码可靠性。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ffeature-engine_feature_engine_5a20507b.png","feature-engine","Feature-engine","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Ffeature-engine_65caeb5f.png","",null,"https:\u002F\u002Fgithub.com\u002Ffeature-engine",[82,86],{"name":83,"color":84,"percentage":85},"Python","#3572A5",99.3,{"name":87,"color":88,"percentage":89},"TeX","#3D6117",0.7,2221,340,"2026-04-04T10:39:36","BSD-3-Clause",1,"Linux, macOS, Windows","未说明",{"notes":98,"python":99,"dependencies":100},"可通过 pip 或 conda 安装；若从源码开发安装，需额外安装测试依赖；支持 Scikit-learn 风格的 fit\u002Ftransform 接口。","3.8+",[101,102,103],"scikit-learn","pandas","numpy",[54,13,51],[106,107,108,101,109,110,111],"machine-learning","data-science","python","feature-engineering","feature-selection","feature-extraction",9,"2026-03-27T02:49:30.150509","2026-04-06T06:54:42.589609",[116,121,126,131,136,141],{"id":117,"question_zh":118,"answer_zh":119,"source_url":120},147,"MeanEncoder 是否支持平滑（smoothing）参数以防止过拟合？","目前 MeanEncoder 尚未实现平滑功能，但该功能已被标记为优先事项，等待贡献者实现。平滑是一种常用技术，用于缓解对稀有或未知类别进行均值编码时的过拟合风险。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F315",{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},148,"CyclicalFeatures 在处理周期性特征（如小时、月份）时如何正确设置周期（period）？","CyclicalFeatures 默认通过最大值估算周期，但这在数据范围不完整（如小时用 0–23 表示）时会导致错误。建议使用 max(n_unique, max_val) 的启发式方法估算周期，并在文档中说明：若训练集未包含完整周期值，用户应手动指定 period 参数以确保正确性。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F765",{"id":127,"question_zh":128,"answer_zh":129,"source_url":130},149,"Feature-Engine 是否支持类型提示（type hints）？","是的，项目已逐步引入类型提示。大部分代码已完成类型注解（见 PR #177），仅 selection 模块尚未完成。建议参考 FastAPI 的做法，通过 mypy 配置逐步启用类型检查，并可结合 pre-commit 提升代码质量。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F152",{"id":132,"question_zh":133,"answer_zh":134,"source_url":135},150,"如何从 datetime 类型变量中提取日期和时间特征？","Feature-Engine 提供了 DatetimeFeatures 变换器（源自 Issue #67 的设计），可统一提取日期（如年、月、日、星期几）和时间（如小时、分钟）特征。最初考虑拆分为三个独立变换器，但最终倾向于使用单一类并通过参数（如 extract_features='date'\u002F'time'\u002F'both'）控制行为，以提升灵活性和易用性。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F67",{"id":137,"question_zh":138,"answer_zh":139,"source_url":140},151,"对于呈对数分布的变量，如何进行指数宽度的离散化（分箱）？","可使用 ExponentialWidthDiscretiser（基于 Issue #445 实现），它按指数尺度（如 0–9, 10–99, 100–999）创建分箱，适用于跨越多个数量级的数据。该方法源自《Feature Engineering for Machine Learning》一书，能更好地保留原始数据的分布特性。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F445",{"id":142,"question_zh":143,"answer_zh":144,"source_url":145},152,"如何将 datetime 列转换为序数（ordinal）数值？","可使用 DatetimeOrdinal 变换器（由 Issue #818 提议并实现），它利用 pandas 的 pd.Timestamp.toordinal() 方法将日期时间转换为整数序数，并支持 inverse_transform 还原。示例代码：X.apply(lambda x: x.map(pd.Timestamp.toordinal, na_action='ignore'))。","https:\u002F\u002Fgithub.com\u002Ffeature-engine\u002Ffeature_engine\u002Fissues\u002F818",[147,152,157,162,167],{"id":148,"version":149,"summary_zh":150,"released_at":151},99829,"v1.2.0","**Big release**\r\n\r\nThe new release adds all of the following:\r\n\r\n- New transformer: datetime\r\n- New transformer: selection by PSI\r\n- New transformer: match variables\r\n- New documentation layout: pydata template\r\n- New documentation organisation: api and user guide\r\n- Sponsorship\r\n- Paper and citation\r\n\r\nAnd a lot more expended functionality of current classes.","2022-01-04T17:09:47",{"id":153,"version":154,"summary_zh":155,"released_at":156},99830,"v1.1.2","Fixes bug in one hot encoder","2021-09-18T17:06:08",{"id":158,"version":159,"summary_zh":160,"released_at":161},99831,"v1.1.1","* Migrates repo to own organisation github.com\u002Ffeature-engine\r\n* adds new transformer (logcp)\r\n* expands transformers with inverse_transform\r\n* expands correlation selectors by allowing callables in the corr method\r\n\r\n","2021-08-06T13:35:56",{"id":163,"version":164,"summary_zh":165,"released_at":166},99832,"1.1.0","- enforce compatibility with Scikit-learn through check_estimator\r\n- new transformer for cyclical features\r\n- add categorical support for variables of type category\r\n- add option to perform categorical transformation to numerical variables\r\n- feature selectors support all cross-validation schemes","2021-06-22T09:56:39",{"id":168,"version":169,"summary_zh":170,"released_at":171},99833,"v1.0.2","In this release we add:\r\n- a new transformer for imputation\r\n- a new transformer to create features\r\n- new jupyter notebooks with demos\r\n- linkd to Kaggle kernels featuring Feature-engine\r\n- fix a number of typos in the documentation\r\n-fix few bugs in the select by target mean transformer.","2021-01-23T12:02:13"]