[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-tslearn-team--tslearn":3,"tool-tslearn-team--tslearn":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",140436,2,"2026-04-05T23:32:43",[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":75,"owner_avatar_url":76,"owner_bio":77,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":77,"owner_website":77,"owner_url":78,"languages":79,"stars":84,"forks":85,"last_commit_at":86,"license":87,"difficulty_score":88,"env_os":89,"env_gpu":90,"env_ram":90,"env_deps":91,"category_tags":96,"github_topics":97,"view_count":23,"oss_zip_url":77,"oss_zip_packed_at":77,"status":16,"created_at":109,"updated_at":110,"faqs":111,"releases":141},3429,"tslearn-team\u002Ftslearn","tslearn","The machine learning toolkit for time series analysis in Python","tslearn 是专为 Python 打造的时间序列机器学习工具箱，旨在降低处理时序数据的门槛。它主要解决了传统机器学习库（如 scikit-learn）难以直接处理时间维度数据、缺乏专用算法以及数据格式转换繁琐等痛点，让开发者能像处理普通表格数据一样轻松应用机器学习模型。\n\n这款工具非常适合数据科学家、研究人员以及需要进行时序分析的 Python 开发者使用。无论是学术探索还是工业界应用，tslearn 都能提供从数据预处理到模型构建的一站式支持。其独特的技术亮点在于原生支持可变长度的时间序列，无需强制对齐或截断数据；同时内置了动态时间规整（DTW）等核心算法，并提供了丰富的实用函数，帮助用户快速将原始数据转换为标准的三维数组格式。此外，它还兼容主流数据集和合成数据生成模块，配合完善的文档与示例，让用户能迅速上手，高效完成分类、聚类和回归等分析任务。","\u003C!-- Our logo and description -->\n\u003Cdiv align=\"center\">\n  \u003Cp>\u003Ca href=\"https:\u002F\u002Ftslearn.readthedocs.io\">\u003Cimage src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Ftslearn-team\u002Ftslearn\u002Fmain\u002Fdocs\u002F_static\u002Ftslearn_logo_white_background.png?cache-control=no-cache\" width=\"20%\" alt=\"tslearn logo\"\u002F>\u003C\u002Fa>\u003C\u002Fp>\n  \u003Ch1>The machine learning toolkit for time series analysis in Python\u003C\u002Fh1>\n\u003C\u002Fdiv>\n\n\u003C!-- The badges -->\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftslearn\">\n        \u003Cimg alt=\"PyPI\" src=\"https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftslearn.svg?cache-control=no-cache\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F\">\n        \u003Cimg alt=\"Python 3.10+\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10+-blue.svg\">\n    \u003C\u002Fa>\n    \u003Ca href=\"http:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002F?badge=stable\">\n        \u003Cimg alt=\"Documentation\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_13d664e1afd7.png\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fdev.azure.com\u002Fromaintavenard\u002Ftslearn\u002F_build\">\n        \u003Cimg alt=\"Build (Azure Pipelines)\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_3323fe13e747.png\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fcodecov.io\u002Fgh\u002Ftslearn-team\u002Ftslearn\">\n        \u003Cimg alt=\"Codecov\" src=\"https:\u002F\u002Fcodecov.io\u002Fgh\u002Ftslearn-team\u002Ftslearn\u002Fbranch\u002Fmain\u002Fgraph\u002Fbadge.svg\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fpepy.tech\u002Fproject\u002Ftslearn\">\n        \u003Cimg alt=\"Downloads\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_f36ee4c7f7f6.png\">\n    \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003C!-- Draw horizontal rule -->\n\u003Chr>\n\n\u003C!-- Table of content -->\n\n| Section | Description |\n|-|-|\n| [Installation](#installation) | Installing the dependencies and tslearn |\n| [Getting started](#getting-started) | A quick introduction on how to use tslearn |\n| [Available features](#available-features) | An extensive overview of tslearn's functionalities |\n| [Documentation](#documentation) | A link to our API reference and a gallery of examples |\n| [Contributing](#contributing) | A guide for heroes willing to contribute |\n| [Citation](#referencing-tslearn) | A citation for tslearn for scholarly articles |\n\n## Installation\u003Ca id=\"installation\">\u003C\u002Fa>\nThere are different alternatives to install tslearn:\n* PyPi: `python -m pip install tslearn`\n* Conda: `conda install -c conda-forge tslearn`\n* Git: `python -m pip install https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Farchive\u002Fmain.zip`\n\nIn order for the installation to be successful, the required dependencies must be installed. For a more detailed guide on how to install tslearn, please see the [Documentation](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002F?badge=stable#installation).\n\n## Getting started\u003Ca id=\"getting-started\">\u003C\u002Fa>\n\n### 1. Getting the data in the right format\ntslearn expects a time series dataset to be formatted as a 3D `numpy` array. The three dimensions correspond to the number of time series, the number of measurements per time series and the number of dimensions respectively (`n_ts, max_sz, d`). In order to get the data in the right format, different solutions exist:\n* [You can use the utility functions such as `to_time_series_dataset`.](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.utils.html#module-tslearn.utils)\n* [You can convert from other popular time series toolkits in Python.](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fintegration_other_software.html)\n* [You can load any of the UCR datasets in the required format.](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.datasets.html#module-tslearn.datasets)\n* [You can generate synthetic data using the `generators` module.](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.generators.html#module-tslearn.generators)\n\nIt should further be noted that tslearn [supports variable-length timeseries](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fvariablelength.html).\n\n```python3\n>>> from tslearn.utils import to_time_series_dataset\n>>> my_first_time_series = [1, 3, 4, 2]\n>>> my_second_time_series = [1, 2, 4, 2]\n>>> my_third_time_series = [1, 2, 4, 2, 2]\n>>> X = to_time_series_dataset([my_first_time_series,\n                                my_second_time_series,\n                                my_third_time_series])\n>>> y = [0, 1, 1]\n```\n\n### 2. Data preprocessing and transformations\nOptionally, tslearn has several utilities to preprocess the data. In order to facilitate the convergence of different algorithms, you can [scale time series](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.preprocessing.html#module-tslearn.preprocessing). Alternatively, in order to speed up training times, one can [resample](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fpreprocessing\u002Ftslearn.preprocessing.TimeSeriesResampler.html#tslearn.preprocessing.TimeSeriesResampler) the data or apply a [piece-wise transformation](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.piecewise.html#module-tslearn.piecewise).\n\n```python3\n>>> from tslearn.preprocessing import TimeSeriesScalerMinMax\n>>> X_scaled = TimeSeriesScalerMinMax().fit_transform(X)\n>>> print(X_scaled)\n[[[0.] [0.667] [1.] [0.333] [nan]]\n [[0.] [0.333] [1.] [0.333] [nan]]\n [[0.] [0.333] [1.] [0.333] [0.333]]]\n```\n\n### 3. Training a model\n\nAfter getting the data in the right format, a model can be trained. Depending on the use case, tslearn supports different tasks: classification, clustering and regression. For an extensive overview of possibilities, check out our [gallery of examples](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Findex.html).\n\n```python3\n>>> from tslearn.neighbors import KNeighborsTimeSeriesClassifier\n>>> knn = KNeighborsTimeSeriesClassifier(n_neighbors=1)\n>>> knn.fit(X_scaled, y)\n>>> print(knn.predict(X_scaled))\n[0 1 1]\n```\n\nAs can be seen, the models in tslearn follow the same API as those of the well-known scikit-learn. Moreover, they are fully compatible with it, allowing to use different scikit-learn utilities such as [hyper-parameter tuning and pipelines](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Fneighbors\u002Fplot_knnts_sklearn.html).\n\n### 4. More analyses\n\ntslearn further allows to perform all different types of analysis. Examples include [calculating barycenters](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.barycenters.html#module-tslearn.barycenters) of a group of time series or calculate the distances between time series using a [variety of distance metrics](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.metrics.html#module-tslearn.metrics).\n\n## Available features\u003Ca id=\"available-features\">\u003C\u002Fa>\n\n| data                                                                                                                                                                                         | processing                                                                                                              | clustering                                                                                                                                                       | classification                                                                                                                                                                          | regression                                                                                                                                                                           | metrics                                                                                                                              |\n|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------|\n| [UCR Datasets](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.datasets.html#module-tslearn.datasets)                                                                           | [Scaling](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.preprocessing.html#module-tslearn.preprocessing) | [TimeSeriesKMeans](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.TimeSeriesKMeans.html#tslearn.clustering.TimeSeriesKMeans) | [KNN Classifier](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fneighbors\u002Ftslearn.neighbors.KNeighborsTimeSeriesClassifier.html#tslearn.neighbors.KNeighborsTimeSeriesClassifier) | [KNN Regressor](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fneighbors\u002Ftslearn.neighbors.KNeighborsTimeSeriesRegressor.html#tslearn.neighbors.KNeighborsTimeSeriesRegressor) | [Dynamic Time Warping](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fmetrics\u002Ftslearn.metrics.dtw.html#tslearn.metrics.dtw)    |\n| [Generators](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.generators.html#module-tslearn.generators)                                                                         | [Piecewise](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.piecewise.html#module-tslearn.piecewise)       | [KShape](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.KShape.html#tslearn.clustering.KShape)                               | [TimeSeriesSVC](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fsvm\u002Ftslearn.svm.TimeSeriesSVC.html#tslearn.svm.TimeSeriesSVC)                                                      | [TimeSeriesSVR](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fsvm\u002Ftslearn.svm.TimeSeriesSVR.html#tslearn.svm.TimeSeriesSVR)                                                   | [Global Alignment Kernel](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fmetrics\u002Ftslearn.metrics.gak.html#tslearn.metrics.gak) |\n| Conversion([1](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.utils.html#module-tslearn.utils), [2](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fintegration_other_software.html)) |                                                                                                                         | [KernelKmeans](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.KernelKMeans.html#tslearn.clustering.KernelKMeans)             | [LearningShapelets](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fshapelets\u002Ftslearn.shapelets.LearningShapelets.html)                                    | [MLP](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.neural_network.html#module-tslearn.neural_network)                                                                | [Barycenters](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.barycenters.html#module-tslearn.barycenters)              |\n|                                                                                                                                                                                              |                                                                                                                         |                                                                                                                                                                  | [Early Classification](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.early_classification.html#module-tslearn.early_classification)                                      |                                                                                                                                                                                      | [Matrix Profile](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.matrix_profile.html#module-tslearn.matrix_profile)     |\n\n\n## Documentation\u003Ca id=\"documentation\">\u003C\u002Fa>\n\nThe documentation is hosted at [readthedocs](http:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Findex.html). It includes an [API](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Freference.html), [gallery of examples](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Findex.html) and a [user guide](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fuser_guide\u002Fuserguide.html).\n\n## Contributing\u003Ca id=\"contributing\">\u003C\u002Fa>\n\nIf you would like to contribute to `tslearn`, please have a look at [our contribution guidelines](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fblob\u002Fmain\u002FCONTRIBUTING.md). A list of interesting TODO's can be found [here](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues?utf8=✓&q=is%3Aissue%20is%3Aopen%20label%3A%22new%20feature%22%20). **If you want other ML methods for time series to be added to this TODO list, do not hesitate to [open an issue](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002Fnew\u002Fchoose)!**\n\n## Referencing tslearn\u003Ca id=\"referencing-tslearn\">\u003C\u002Fa>\n\nIf you use `tslearn` in a scientific publication, we would appreciate citations:\n\n```bibtex\n@article{JMLR:v21:20-091,\n  author  = {Romain Tavenard and Johann Faouzi and Gilles Vandewiele and \n             Felix Divo and Guillaume Androz and Chester Holtz and \n             Marie Payne and Roman Yurchak and Marc Ru{\\ss}wurm and \n             Kushal Kolar and Eli Woods},\n  title   = {Tslearn, A Machine Learning Toolkit for Time Series Data},\n  journal = {Journal of Machine Learning Research},\n  year    = {2020},\n  volume  = {21},\n  number  = {118},\n  pages   = {1-6},\n  url     = {http:\u002F\u002Fjmlr.org\u002Fpapers\u002Fv21\u002F20-091.html}\n}\n```\n\n#### Acknowledgments\nAuthors would like to thank Mathieu Blondel for providing code for [Kernel k-means](https:\u002F\u002Fgist.github.com\u002Fmblondel\u002F6230787) and [Soft-DTW](https:\u002F\u002Fgithub.com\u002Fmblondel\u002Fsoft-dtw), and to Mehran Maghoumi for his [`torch`-compatible implementation of SoftDTW](https:\u002F\u002Fgithub.com\u002FMaghoumi\u002Fpytorch-softdtw-cuda).\n","\u003C!-- 我们的Logo和简介 -->\n\u003Cdiv align=\"center\">\n  \u003Cp>\u003Ca href=\"https:\u002F\u002Ftslearn.readthedocs.io\">\u003Cimage src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Ftslearn-team\u002Ftslearn\u002Fmain\u002Fdocs\u002F_static\u002Ftslearn_logo_white_background.png?cache-control=no-cache\" width=\"20%\" alt=\"tslearn logo\"\u002F>\u003C\u002Fa>\u003C\u002Fp>\n  \u003Ch1>Python中用于时间序列分析的机器学习工具包\u003C\u002Fh1>\n\u003C\u002Fdiv>\n\n\u003C!-- 各种徽章 -->\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftslearn\">\n        \u003Cimg alt=\"PyPI\" src=\"https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftslearn.svg?cache-control=no-cache\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F\">\n        \u003Cimg alt=\"Python 3.10+\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10+-blue.svg\">\n    \u003C\u002Fa>\n    \u003Ca href=\"http:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002F?badge=stable\">\n        \u003Cimg alt=\"文档\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_13d664e1afd7.png\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fdev.azure.com\u002Fromaintavenard\u002Ftslearn\u002F_build\">\n        \u003Cimg alt=\"构建（Azure管道）\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_3323fe13e747.png\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fcodecov.io\u002Fgh\u002Ftslearn-team\u002Ftslearn\">\n        \u003Cimg alt=\"Codecov\" src=\"https:\u002F\u002Fcodecov.io\u002Fgh\u002Ftslearn-team\u002Ftslearn\u002Fbranch\u002Fmain\u002Fgraph\u002Fbadge.svg\">\n    \u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fpepy.tech\u002Fproject\u002Ftslearn\">\n        \u003Cimg alt=\"下载量\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_readme_f36ee4c7f7f6.png\">\n    \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003C!-- 绘制水平线 -->\n\u003Chr>\n\n\u003C!-- 目录 -->\n\n| 章节 | 描述 |\n|-|-|\n| [安装](#installation) | 安装依赖项和tslearn |\n| [入门](#getting-started) | tslearn的快速使用介绍 |\n| [可用功能](#available-features) | tslearn功能的全面概述 |\n| [文档](#documentation) | 我们的API参考链接和示例图库 |\n| [贡献](#contributing) | 针对有意贡献的开发者的指南 |\n| [引用](#referencing-tslearn) | 学术文章中引用tslearn的方式 |\n\n## 安装\u003Ca id=\"installation\">\u003C\u002Fa>\n安装tslearn有多种方式：\n* PyPi：`python -m pip install tslearn`\n* Conda：`conda install -c conda-forge tslearn`\n* Git：`python -m pip install https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Farchive\u002Fmain.zip`\n\n为确保安装成功，必须先安装所需的依赖项。有关如何安装tslearn的更详细指南，请参阅[文档](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002F?badge=stable#installation)。\n\n## 入门\u003Ca id=\"getting-started\">\u003C\u002Fa>\n\n### 1. 将数据整理成正确格式\ntslearn期望时间序列数据集以三维`numpy`数组的形式呈现。这三个维度分别对应时间序列的数量、每个时间序列的测量点数以及每个时间序列的维度数（`n_ts, max_sz, d`）。为了将数据整理成正确格式，有以下几种方法：\n* [可以使用诸如`to_time_series_dataset`之类的实用函数。](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.utils.html#module-tslearn.utils)\n* [可以从其他流行的Python时间序列工具包中进行转换。](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fintegration_other_software.html)\n* [可以直接加载UCR数据集中符合要求的数据集。](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.datasets.html#module-tslearn.datasets)\n* [可以使用`generators`模块生成合成数据。](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.generators.html#module-tslearn.generators)\n\n此外，需要注意的是，tslearn还[支持变长时间序列](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fvariablelength.html)。\n\n```python3\n>>> from tslearn.utils import to_time_series_dataset\n>>> my_first_time_series = [1, 3, 4, 2]\n>>> my_second_time_series = [1, 2, 4, 2]\n>>> my_third_time_series = [1, 2, 4, 2, 2]\n>>> X = to_time_series_dataset([my_first_time_series,\n                                my_second_time_series,\n                                my_third_time_series])\n>>> y = [0, 1, 1]\n```\n\n### 2. 数据预处理与变换\n可选地，tslearn提供了一些用于数据预处理的工具。为了促进不同算法的收敛，可以对时间序列进行[缩放](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.preprocessing.html#module-tslearn.preprocessing)。或者，为了加快训练速度，还可以对数据进行[重采样](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fpreprocessing\u002Ftslearn.preprocessing.TimeSeriesResampler.html#tslearn.preprocessing.TimeSeriesResampler)或应用[分段变换](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.piecewise.html#module-tslearn.piecewise)。\n\n```python3\n>>> from tslearn.preprocessing import TimeSeriesScalerMinMax\n>>> X_scaled = TimeSeriesScalerMinMax().fit_transform(X)\n>>> print(X_scaled)\n[[[0.] [0.667] [1.] [0.333] [nan]]\n [[0.] [0.333] [1.] [0.333] [nan]]\n [[0.] [0.333] [1.] [0.333] [0.333]]]\n```\n\n### 3. 训练模型\n\n在将数据整理成正确格式后，就可以开始训练模型了。根据具体的应用场景，tslearn支持分类、聚类和回归等任务。如需了解更全面的可能性，请查看我们的[示例图库](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Findex.html)。\n\n```python3\n>>> from tslearn.neighbors import KNeighborsTimeSeriesClassifier\n>>> knn = KNeighborsTimeSeriesClassifier(n_neighbors=1)\n>>> knn.fit(X_scaled, y)\n>>> print(knn.predict(X_scaled))\n[0 1 1]\n```\n\n可以看出，tslearn中的模型遵循与知名scikit-learn相同的API。此外，它们与scikit-learn完全兼容，因此可以使用scikit-learn的各种工具，例如[超参数调优和流水线](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Fneighbors\u002Fplot_knnts_sklearn.html)。\n\n### 4. 更多分析\n\ntslearn还允许执行各种类型的分析。例如，可以计算一组时间序列的[质心](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.barycenters.html#module-tslearn.barycenters)，或者使用[多种距离度量](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.metrics.html#module-tslearn.metrics)来计算时间序列之间的距离。\n\n## 可用功能\u003Ca id=\"available-features\">\u003C\u002Fa>\n\n| 数据                                                                                                                                                                                         | 处理                                                                                                              | 聚类                                                                                                                                                       | 分类                                                                                                                                                                          | 回归                                                                                                                                                                           | 度量方法                                                                                                                              |\n|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------|\n| [UCR 数据集](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.datasets.html#module-tslearn.datasets)                                                                           | [缩放](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.preprocessing.html#module-tslearn.preprocessing) | [TimeSeriesKMeans](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.TimeSeriesKMeans.html#tslearn.clustering.TimeSeriesKMeans) | [KNN 分类器](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fneighbors\u002Ftslearn.neighbors.KNeighborsTimeSeriesClassifier.html#tslearn.neighbors.KNeighborsTimeSeriesClassifier) | [KNN 回归器](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fneighbors\u002Ftslearn.neighbors.KNeighborsTimeSeriesRegressor.html#tslearn.neighbors.KNeighborsTimeSeriesRegressor) | [动态时间规整](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fmetrics\u002Ftslearn.metrics.dtw.html#tslearn.metrics.dtw)    |\n| [生成器](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.generators.html#module-tslearn.generators)                                                                         | [分段](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.piecewise.html#module-tslearn.piecewise)       | [KShape](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.KShape.html#tslearn.clustering.KShape)                               | [TimeSeriesSVC](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fsvm\u002Ftslearn.svm.TimeSeriesSVC.html#tslearn.svm.TimeSeriesSVC)                                                      | [TimeSeriesSVR](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fsvm\u002Ftslearn.svm.TimeSeriesSVR.html#tslearn.svm.TimeSeriesSVR)                                                   | [全局对齐核](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fmetrics\u002Ftslearn.metrics.gak.html#tslearn.metrics.gak) |\n| 转换([1](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.utils.html#module-tslearn.utils), [2](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fintegration_other_software.html)) |                                                                                                                         | [KernelKmeans](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fclustering\u002Ftslearn.clustering.KernelKMeans.html#tslearn.clustering.KernelKMeans)             | [LearningShapelets](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Fshapelets\u002Ftslearn.shapelets.LearningShapelets.html)                                    | [MLP](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.neural_network.html#module-tslearn.neural_network)                                                                | [质心](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.barycenters.html#module-tslearn.barycenters)              |\n|                                                                                                                                                                                              |                                                                                                                         |                                                                                                                                                                  | [早期分类](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.early_classification.html#module-tslearn.early_classification)                                      |                                                                                                                                                                                      | [矩阵概况](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fgen_modules\u002Ftslearn.matrix_profile.html#module-tslearn.matrix_profile)     |\n\n\n## 文档\u003Ca id=\"documentation\">\u003C\u002Fa>\n\n文档托管在 [readthedocs](http:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Findex.html) 上。它包括 [API](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Freference.html)、[示例图库](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fauto_examples\u002Findex.html) 和 [用户指南](https:\u002F\u002Ftslearn.readthedocs.io\u002Fen\u002Fstable\u002Fuser_guide\u002Fuserguide.html)。\n\n## 贡献\u003Ca id=\"contributing\">\u003C\u002Fa>\n\n如果您希望为 `tslearn` 做出贡献，请查看我们的 [贡献指南](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fblob\u002Fmain\u002FCONTRIBUTING.md)。有趣待办事项列表可以在这里找到 [这里](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues?utf8=✓&q=is%3Aissue%20is%3Aopen%20label%3A%22new%20feature%22%20)。**如果您希望将其他针对时间序列的机器学习方法添加到此待办事项列表中，请随时 [提交一个问题](https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002Fnew\u002Fchoose)!**\n\n## 引用 tslearn\u003Ca id=\"referencing-tslearn\">\u003C\u002Fa>\n\n如果您在科学出版物中使用 `tslearn`，我们非常感谢您能进行引用：\n\n```bibtex\n@article{JMLR:v21:20-091,\n  author  = {Romain Tavenard 和 Johann Faouzi 和 Gilles Vandewiele 和 \n             Felix Divo 和 Guillaume Androz 和 Chester Holtz 和 \n             Marie Payne 和 Roman Yurchak 和 Marc Ru{\\ss}wurm 和 \n             Kushal Kolar 和 Eli Woods},\n  title   = {Tslearn，一个用于时间序列数据的机器学习工具包},\n  journal = {机器学习研究期刊},\n  year    = {2020},\n  volume  = {21},\n  number  = {118},\n  pages   = {1-6},\n  url     = {http:\u002F\u002Fjmlr.org\u002Fpapers\u002Fv21\u002F20-091.html}\n}\n```\n\n#### 致谢\n作者谨向 Mathieu Blondel 表示感谢，感谢他提供了 [Kernel k-means](https:\u002F\u002Fgist.github.com\u002Fmblondel\u002F6230787) 和 [Soft-DTW](https:\u002F\u002Fgithub.com\u002Fmblondel\u002Fsoft-dtw) 的代码；同时也感谢 Mehran Maghoumi 提供了与 `torch` 兼容的 SoftDTW 实现 [pytorch-softdtw-cuda](https:\u002F\u002Fgithub.com\u002FMaghoumi\u002Fpytorch-softdtw-cuda)。","# tslearn 快速上手指南\n\ntslearn 是一个专为 Python 设计的时序分析机器学习工具包，提供了丰富的时序数据处理、聚类、分类和回归算法，并完美兼容 scikit-learn 接口。\n\n## 环境准备\n\n*   **操作系统**：支持 Windows、macOS 和 Linux。\n*   **Python 版本**：需要 Python 3.10 或更高版本。\n*   **前置依赖**：安装时会自动处理核心依赖（如 `numpy`, `scipy`, `scikit-learn` 等），无需手动预装。\n\n## 安装步骤\n\n推荐使用国内镜像源以加速下载。你可以选择以下任意一种方式进行安装：\n\n### 方式一：使用 pip 安装（推荐）\n```bash\npython -m pip install tslearn -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n### 方式二：使用 Conda 安装\n如果你使用 Anaconda 或 Miniconda 环境：\n```bash\nconda install -c conda-forge tslearn\n```\n\n### 方式三：从源码安装（最新开发版）\n```bash\npython -m pip install https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Farchive\u002Fmain.zip\n```\n\n## 基本使用\n\ntslearn 的使用流程与 scikit-learn 高度一致，主要分为数据格式化、预处理和模型训练三个步骤。\n\n### 1. 数据格式化\ntslearn 要求输入数据为三维 `numpy` 数组，形状为 `(样本数, 时间步长, 特征维度)`。它原生支持变长时间序列。\n\n```python3\nfrom tslearn.utils import to_time_series_dataset\n\n# 原始变长数据\nmy_first_time_series = [1, 3, 4, 2]\nmy_second_time_series = [1, 2, 4, 2]\nmy_third_time_series = [1, 2, 4, 2, 2]\n\n# 转换为 tslearn 所需的格式 (自动填充 NaN 以对齐长度)\nX = to_time_series_dataset([my_first_time_series,\n                            my_second_time_series,\n                            my_third_time_series])\ny = [0, 1, 1]  # 标签\n```\n\n### 2. 数据预处理\n为了提高模型收敛速度和效果，通常需要对时序数据进行缩放（Scaling）。\n\n```python3\nfrom tslearn.preprocessing import TimeSeriesScalerMinMax\n\n# 使用最小 - 最大归一化\nscaler = TimeSeriesScalerMinMax()\nX_scaled = scaler.fit_transform(X)\n```\n\n### 3. 模型训练与预测\n以下示例展示了如何使用 KNN 进行时序分类，其 API 风格与 scikit-learn 完全相同。\n\n```python3\nfrom tslearn.neighbors import KNeighborsTimeSeriesClassifier\n\n# 初始化模型\nknn = KNeighborsTimeSeriesClassifier(n_neighbors=1)\n\n# 训练模型\nknn.fit(X_scaled, y)\n\n# 进行预测\npredictions = knn.predict(X_scaled)\nprint(predictions)\n# 输出: [0 1 1]\n```\n\n> **提示**：由于兼容 scikit-learn，你可以直接将 tslearn 的模型放入 `sklearn.pipeline.Pipeline` 中，或使用 `sklearn.model_selection.GridSearchCV` 进行超参数调优。","某工业预测性维护团队需要分析数千台旋转电机的振动传感器数据，以提前识别设备故障模式。\n\n### 没有 tslearn 时\n- **数据格式混乱**：不同电机运行时长不一，导致采集到的时间序列长度参差不齐，团队需编写大量自定义代码进行填充或截断，极易丢失关键波形特征。\n- **算法实现困难**：想要使用动态时间规整（DTW）来计算两条振动曲线的相似度，必须从零实现复杂算法或寻找不稳定的第三方脚本，开发周期长且易出错。\n- **模型适配繁琐**：标准的 Scikit-learn 分类器无法直接处理三维时间序列数组，工程师不得不手动展平数据或设计复杂的预处理管道，导致模型准确率大幅下降。\n- **缺乏专用工具**：面对变长序列聚类需求，现有通用机器学习库束手无策，团队只能放弃高级分析，退回到简单的阈值报警，漏报率居高不下。\n\n### 使用 tslearn 后\n- **一键格式统一**：利用 `to_time_series_dataset` 工具函数，轻松将不等长的原始振动数据转换为标准的三维 NumPy 数组，原生支持变长序列，完整保留故障前的细微征兆。\n- **内置核心算法**：直接调用封装好的 DTW 距离度量与 K-Shape 聚类算法，几行代码即可实现高精度的相似故障模式匹配，无需重复造轮子。\n- **无缝对接生态**：tslearn 提供的分类器和聚类器完全兼容 Scikit-learn 接口，团队可直接复用现有的模型评估与调参流程，快速部署高可用性的故障预测模型。\n- **深度分析赋能**：借助丰富的时序专属功能，成功从噪声中分离出早期磨损特征，将故障预警时间从“事后”提前到“事前”，显著降低停机损失。\n\ntslearn 让 Python 开发者能够像处理表格数据一样简单高效地挖掘时间序列价值，彻底打破了时序分析的高门槛。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Ftslearn-team_tslearn_504a86fc.png","tslearn-team","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Ftslearn-team_ad238029.png",null,"https:\u002F\u002Fgithub.com\u002Ftslearn-team",[80],{"name":81,"color":82,"percentage":83},"Python","#3572A5",100,3136,368,"2026-04-02T08:43:41","BSD-2-Clause",1,"","未说明",{"notes":92,"python":93,"dependencies":94},"该工具是用于时间序列分析的机器学习库，兼容 scikit-learn API。支持通过 PyPI、Conda 或 Git 安装。数据需格式化为 3D numpy 数组，支持可变长度时间序列。","3.10+",[95],"numpy",[13],[98,99,100,101,102,103,104,105,106,107,108],"time-series","timeseries","time-series-analysis","time-series-clustering","machine-learning","machine-learning-algorithms","machinelearning","dtw","python","time-series-classification","dynamic-time-warping","2026-03-27T02:49:30.150509","2026-04-06T09:43:40.958923",[112,117,122,127,132,137],{"id":113,"question_zh":114,"answer_zh":115,"source_url":116},15736,"安装 tslearn 时遇到 'Could not build wheels for numpy' 错误怎么办？","该错误通常由 NumPy 版本不兼容引起。解决方案包括：\n1. 升级依赖：项目已将 NumPy 最低版本要求提升至 1.24.3，请确保环境满足此要求。\n2. 使用旧版 Python：尝试使用 Python 3.10 等较稳定版本。\n3. 手动安装 Wheel：下载对应版本的 numpy wheel 文件（例如 numpy-1.22.0-cp310-cp310-win_amd64.whl）并在虚拟环境中通过 `pip install \"文件名.whl\"` 进行安装。\n4. 清理环境：如果是 ROS\u002FCMake 相关报错，尝试删除 build 和 devel 文件夹，退出 conda 环境 (`conda deactivate`) 后重新编译。","https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F399",{"id":118,"question_zh":119,"answer_zh":120,"source_url":121},15737,"如何在 tslearn 中启用并行计算以加速算法（如 k-Shape）？","从 tslearn v0.2 版本开始，大多数估算器（estimators）以及距离计算函数（如 `cdist_dtw`, `cdist_gak`）都支持并行计算。您只需在初始化估算器或调用函数时指定 `n_jobs` 参数即可。例如：`model = KShape(n_jobs=-1)` 将使用所有可用的 CPU 核心，从而显著减少处理大规模数据的时间。","https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F55",{"id":123,"question_zh":124,"answer_zh":125,"source_url":126},15738,"tslearn 是否支持多平台二进制安装包，以避免本地编译？","是的，tslearn 现已支持为所有主要平台构建二进制 wheels。维护者使用了 `cibuildwheel` 工具来自动化构建过程。这意味着用户现在可以直接通过 `pip install tslearn` 进行安装，而无需在本地安装编译器或 Cython。该方案已覆盖 Windows、macOS 和 Linux，并支持 Python 2.7 及 3.5 到 3.8 等多个版本。","https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F150",{"id":128,"question_zh":129,"answer_zh":130,"source_url":131},15739,"在使用基于 SAX+MINDIST 的 kNN 分类器时，可以使用哪些距离度量（metrics）？","除了默认的 DTW 之外，tslearn 已经实现了对其他度量的支持。根据开发进度（v0.4 版本起），您可以使用欧几里得距离（euclidean）或专门针对 SAX 表示的距离度量。具体可用的参数取决于您使用的类版本，建议查阅最新文档或源码确认当前支持的 metric 列表。","https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F28",{"id":133,"question_zh":134,"answer_zh":135,"source_url":136},15740,"tslearn 中的多维时间序列数据格式是怎样的？与 STUMPY 有何不同？","在 tslearn 中，一组多维时间序列的形状通常为 `(n_series, n_timestamps, n_features)`，其中 `n_series` 是序列数量，`n_timestamps` 是时间步长，`n_features` 是维度数。单个多维时间序列的形状为 `(n_timestamps, n_features)`。\n相比之下，STUMPY 库中多维时间序列的形状通常定义为 `(n_features, n_timestamps)`。在进行跨库迁移或对比时，需要注意这种维度顺序的差异（即是否需要转置数据）。","https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F260",{"id":138,"question_zh":139,"answer_zh":140,"source_url":136},15741,"tslearn 是否计划集成更高效的 Matrix Profile 算法（如 STUMPY）？","是的，社区正在讨论将 STUMPY 作为可选依赖集成到 tslearn 的 MatrixProfile 模块中，以替代现有的朴素实现，从而获得更好的可扩展性。未来的实现计划允许用户在初始化时通过参数选择具体的后端实现（原生或 STUMPY），并在 `_transform` 方法中自动调用相应的函数。这将特别有利于处理长序列或多维序列的场景。",[142,147,152,157,162,166,171,176,181,185,189,194,198,202,206,210,214,218,222,226],{"id":143,"version":144,"summary_zh":145,"released_at":146},90426,"v0.8.1","v0.8.1\n\n* `TimeSeriesScalerMinMax` 和 `TimeSeriesScalerMeanVariance` 现在支持可变长度的时间序列。\n* 提升 Numba 依赖的最低版本。\n* 修复 Shapelets 定位中对可变长度时间序列的支持。","2026-03-13T13:04:55",{"id":148,"version":149,"summary_zh":150,"released_at":151},90427,"v0.8.0","版本 0.8.0\n\n变更\n\n    重构 cdist_dtw 及其底层的 Numba 加速 DTW 计算，以提升 numpy 后端的运行速度。\n    compute_mask、sakoe_chiba_mask 和 itakura_mask 返回的掩码现在使用布尔值。\n    后端属性现为动态属性。\n    确保在 to_time_series_dataset 中保留数据类型。\n    修复了 K-shape 算法中多变量时间序列的质心计算问题（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F288）。\n    修复了 LearningShapelet 的 JSON 和 Pickle 序列化问题（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F387）（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F403）。\n\n新增\n\n    针对时间序列的 DBSCAN 估计器（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F598）","2026-02-19T08:54:37",{"id":153,"version":154,"summary_zh":155,"released_at":156},90428,"v0.7.0","版本 0.7.0\n\n变更\n\n-    当使用全局对齐核且 sigma 接近零时，会抛出明确的异常（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F440）\n-    修复了 K-shape 形状提取过程中的偏移问题（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F385）\n-    支持至 scikit-learn 1.7 版本（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F549）\n-    修复了针对可变长度时间序列的 LearningShapelets 算法（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F352）\n-    shapelets 模块现依赖于 Keras 3 及以上版本，并可通过 KERAS_BACKEND 环境变量选择底层后端。默认按 torch、tensorflow、jax 的顺序，优先使用首个已安装的后端。\n\n移除\n\n-    停止对 Python 3.8 和 3.9 版本的支持\n\n新增\n\n-   为最小-最大归一化和均值-方差标准化添加了 per_timeseries 和 per_feature 选项（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F536）\n-   TimeSeriesImputer 类：用于时间序列的缺失值插补器（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F564）\n-    集成了 Fréchet 距离度量与 K 近邻算法（https:\u002F\u002Fgithub.com\u002Ftslearn-team\u002Ftslearn\u002Fissues\u002F402）","2025-11-13T07:49:46",{"id":158,"version":159,"summary_zh":160,"released_at":161},90429,"v0.6.4","版本 0.6.4\n\n与 scikit-learn \u003C 1.7 兼容","2025-07-02T12:20:58",{"id":163,"version":164,"summary_zh":77,"released_at":165},90430,"v0.6.3","2023-12-12T14:35:12",{"id":167,"version":168,"summary_zh":169,"released_at":170},90431,"v0.6.2","# 本版本修复的 bug\n\n* 修复了在提供 `dataset2` 时，`cdist_soft_dtw_normalized` 的归一化项计算错误的问题。\n* 修复了 UCR\u002FUEA 数据集的下载链接。","2023-08-21T13:24:24",{"id":172,"version":173,"summary_zh":174,"released_at":175},90432,"v0.6.1","版本 0.6.1","2023-07-05T12:11:29",{"id":177,"version":178,"summary_zh":179,"released_at":180},90433,"v0.6.0","版本 0.6.0","2023-07-03T19:34:15",{"id":182,"version":183,"summary_zh":77,"released_at":184},90434,"v0.5.3.2","2023-01-20T13:21:39",{"id":186,"version":187,"summary_zh":77,"released_at":188},90435,"v0.5.3.1","2023-01-20T13:05:03",{"id":190,"version":191,"summary_zh":192,"released_at":193},90436,"v0.5.3","版本 0.5.3","2023-01-18T14:02:42",{"id":195,"version":196,"summary_zh":77,"released_at":197},90437,"v0.5.2","2021-08-16T07:09:52",{"id":199,"version":200,"summary_zh":77,"released_at":201},90438,"v0.5.1.0","2021-05-17T21:36:57",{"id":203,"version":204,"summary_zh":77,"released_at":205},90439,"v0.5.1","2021-05-17T21:19:10",{"id":207,"version":208,"summary_zh":77,"released_at":209},90440,"v0.5.0.5","2021-01-25T15:19:11",{"id":211,"version":212,"summary_zh":77,"released_at":213},90441,"v0.5.0.3","2021-01-25T14:32:00",{"id":215,"version":216,"summary_zh":77,"released_at":217},90442,"v0.5.0.2","2021-01-24T21:17:55",{"id":219,"version":220,"summary_zh":77,"released_at":221},90443,"v0.5.0.1","2021-01-24T21:05:22",{"id":223,"version":224,"summary_zh":77,"released_at":225},90444,"v0.5.0","2021-01-24T20:42:25",{"id":227,"version":228,"summary_zh":77,"released_at":229},90445,"v0.4.1","2020-06-18T15:54:49"]