[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118835-en":3,"doc-seo-118835-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},118835,34359740700684,"Finn","https://ap-avatar.wpscdn.com/avatar/1f400023980c374ae676?_k=1777273430885731487",6,"Technology","Non-Sequential Machine Learning Pipelines with pyWATTS - deRSE23-Conference","This presentation introduces non-sequential machine learning pipelines implemented with pyWATTS, focusing on workflow design for time-series tasks such as forecasting and classification. It explains how pipelines can be structured as directed acyclic graphs to support non-linear dependency between preprocessing, feature extraction, and multiple model components. The talk contrasts benefits like simpler hyperparameter tuning, easier pipeline persistence, and a single fit call for end-to-end training. It also presents pyWATTS’s three pipeline construction APIs—functional, imperative, and constructor—then demonstrates a functional API example for combining calendar features, lag features, and sklearn-based forecasting components.","Non-Sequential Machine Learning Pipelines with pyWATTS  \ndeRSE23-Conference for Research Software Engineering in Germany  \nBenedikt Heidrich, Kaleb Phipps,  \nStefan Meisenbacher, Marian Turowski, Oliver Neumann, Ralf Mikut, Veit Hagenmeyer  \nKIT – The Research University in the Helmholtz Association [www.kit.edu](www.kit.edu)  \nMachine Learning  \nPipelines  \nInput  \nPre-Processing  \nTask (Forecasting, Classification)  \nOutput  \n2 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \nMachine Learning  \nPipelines  \n| Input |  | Pre-Processing |  | Task (Forecasting, Classification) |  | Output |\n| --- | --- | --- | --- | --- | --- | --- |\n|  |  |  |  |  |  |  |\n\n3 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \nMachine Learning  \nPipelines  \n| Pipeline\u003Cbr>\u003Cbr>Input\u003Cbr>\u003Cbr>~~ ~~ Pre-Processing |  |  |  |  |  |\n| --- | --- | --- | --- | --- | --- |\n|  |  | Task (Forecasting, Classification) |  | Output |  |\n|  |  |  |  |  |  |\n|  |  |  |  |  |  |\n\n4 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \nMachine Learning  \nPipelines  \n| Pipeline\u003Cbr>\u003Cbr>Input\u003Cbr>\u003Cbr>~~ ~~ Pre-Processing |  |  |  |  |  |\n| --- | --- | --- | --- | --- | --- |\n|  |  | Task (Forecasting, Classification) |  | Output |  |\n|  |  |  |  |  |  |\n|  |  |  |  |  |  |\n\n● Why is this beneficial?  \n○ Hyperparameter tuning is simpler-you can tune the whole pipeline.  \n○ It is easy to handle-there is only one pipeline object that needs to be saved.  \n○ You only have to call the fit method once to train the entire pipeline.  \n5 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \nMachine Learning  \nPipelines  \n| Pipeline\u003Cbr>\u003Cbr>Input\u003Cbr>\u003Cbr>~~ ~~ Pre-Processing |  |  |  |  |  |\n| --- | --- | --- | --- | --- | --- |\n|  |  | Task (Forecasting, Classification) |  | Output |  |\n|  |  |  |  |  |  |\n|  |  |  |  |  |  |\n\n● Why is this beneficial?  \n○ Hyperparameter tuning is simpler-you can tune the whole pipeline.  \n○ It is easy to handle-there is only one pipeline object that needs to be saved.  \n○ You only have to call the fit method once to train the entire pipeline.  \n● Existing sequential pipeline implementations in…  \n○ sklearn  \n○ sktime  \n6 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \nHowever: Many use cases are non-sequential  \nFor example, electricity price forecasting  \nPrice  \nTime Series Index  \nLoad  \nStandard Scaler  \nPrice Forecaster  \nInverse Scaling  \nStandard Scaler  \nStandard Scaler  \nInverse Scaling  \nCalendar Features  \nLoad Forecaster  \n7 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \npyWATTS: Python Workflow Automation Tool for Time Series  \n● pyWATTS models pipeline as directed acyclic graphs enabling non-sequential workflows:  \n○ Code is easier to write and intuitive to understand.  \n○ Hyperparameter optimisation is easier.  \n○ Possible to combine multiple tasks in one pipeline.  \n● pyWATTS provides three different APIs for creating pipelines:  \n○ Functional API  \n○ Imperative API  \n○ Constructor API  \n8 Heidrich et al. deRSE23-Non-Sequential Machine Learning Pipelines with pyWATTS Institute for Automation and Applied Informatics (IAI)  \npyWATTS: Python Workflow Automation Tool for Time Series  \n● pyWATTS models pipeline as directed acyclic graphs enabling non-sequential workflows:  \n○ Code is easier to write and intuitive to understand.  \n○ Hyperparameter optimisation is easier.  \n○ Possible to combine multiple tasks in one pipeline.  \n● pyWATTS provides three different APIs for creating pipelines:  \n○ Functional API  \n○ Imperative API  \n○ Co","cbCaiqpSu4WUNApr","https://ap.wps.com/l/cbCaiqpSu4WUNApr","pdf",461284,1,25,"English","en",105,"# Motivation and Pipeline Structure\n## Input, Pre-Processing, Task, and Output\n## Why Non-Sequential Pipelines Help\n# Non-Sequential Use Cases\n## Electricity Price Forecasting Example\n# pyWATTS Overview\n## Directed Acyclic Graph Workflows\n## APIs for Creating Pipelines\n# Functional API Example\n## Building and Training the Pipeline","[{\"question\":\"What is the main idea behind non-sequential machine learning pipelines with pyWATTS?\",\"answer\":\"pyWATTS models pipelines as directed acyclic graphs so workflows can include non-sequential dependencies between preprocessing and multiple tasks. This enables easier composition of time-series operations in one pipeline.\"},{\"question\":\"Why is end-to-end training beneficial in this pipeline approach?\",\"answer\":\"Hyperparameter tuning becomes simpler because the whole pipeline can be tuned together. Handling is easier since there is only one pipeline object to save, and training can be done by calling fit once.\"},{\"question\":\"Which APIs does pyWATTS provide for creating pipelines?\",\"answer\":\"pyWATTS provides three APIs: Functional API, Imperative API, and Constructor API. The presentation specifically focuses on the Functional API with a code example.\"}]","Non-Sequential Machine Learning Pipelines with pyWATTS - deRSE23-Conference | PDF",1785720522,63,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"non-sequential-machine-learning-pipelines-with-pywatts-derse23-conference","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/technology/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/non-sequential-machine-learning-pipelines-with-pywatts-derse23-conference/118835/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What is the main idea behind non-sequential machine learning pipelines with pyWATTS?","Question",{"text":75,"@type":76},"pyWATTS models pipelines as directed acyclic graphs so workflows can include non-sequential dependencies between preprocessing and multiple tasks. This enables easier composition of time-series operations in one pipeline.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why is end-to-end training beneficial in this pipeline approach?",{"text":80,"@type":76},"Hyperparameter tuning becomes simpler because the whole pipeline can be tuned together. Handling is easier since there is only one pipeline object to save, and training can be done by calling fit once.",{"name":82,"@type":73,"acceptedAnswer":83},"Which APIs does pyWATTS provide for creating pipelines?",{"text":84,"@type":76},"pyWATTS provides three APIs: Functional API, Imperative API, and Constructor API. The presentation specifically focuses on the Functional API with a code example.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":46,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":46,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]