{"categories":[{"label":"Artificial Intelligence","url":"https://skillfed.io/packages/category/scientific-engineering-artificial-intelligence/9"},{"label":"Application Frameworks","url":"https://skillfed.io/packages/category/software-development-libraries-application-frameworks/5"}],"enrichment":{"capability":"PyPOTS provides machine learning algorithms for time series with missing values, supporting imputation, classification, clustering, forecasting, and anomaly detection on partially-observed multivariate data.","skillfed_tags":["time-series","missing-data","neural-networks"],"use_cases":["Impute missing sensor readings in IoT or industrial monitoring data before downstream analysis","Forecast equipment failures or system behavior when historical data has communication gaps","Classify time series patterns (e.g., patient conditions) despite incomplete medical records","Detect anomalies in multivariate time series with irregular sampling or missing observations","Cluster similar time series trajectories when some features are not always recorded"],"what_it_does":"PyPOTS is a machine learning toolkit designed specifically for time series data with missing values\u2014a common real-world problem caused by sensor failures, communication errors, or data collection issues. It fills a gap by providing a unified interface to classical and state-of-the-art algorithms for handling partially-observed time series, rather than requiring researchers to build custom solutions or adapt general-purpose libraries.\n\nThe package supports five core tasks: imputation (filling missing values), forecasting, classification, clustering, and anomaly detection. It includes neural network models, transformers adapted for incomplete data, and LLM-based approaches. All models share consistent APIs and come with documentation and examples. The toolkit depends on torch, transformers, scikit-learn, and other standard data science libraries, making it suitable for researchers and engineers working with real-world time series that contain gaps.","worth_installing":"Yes, if you work with incomplete time series data and need a unified toolkit with multiple algorithms. The package is actively maintained, BSD-licensed, and has no known vulnerabilities. The 16 runtime dependencies (especially torch and transformers) add significant install size and complexity; evaluate whether you need the full suite or can use a lighter alternative if you only need imputation or forecasting."},"id":"pypots","links":{"html":"https://skillfed.io/packages/pypots","md":"https://skillfed.io/packages/pypots.md","pypi":"https://pypi.org/project/pypots/"},"maintenance":{"status":"active"},"meta":{"latest_release":"2026-05-05","license_spdx":null,"license_treatment":"permissive","name":"pypots","python_support":"supports_current","summary":"A Python Toolbox for Machine Learning on Partially-Observed Time Series"},"popularity":{"monthly_downloads":123029,"position":11924,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"1.5"}