alan-turing-institute/sktime

Features

• Time Series Forecasting - Provides a unified framework for predicting future values in temporal data sequences using diverse learning algorithms.
• Time Series Machine Learning Frameworks - A unified interface for applying machine learning algorithms to time series data through standardized workflows.
• Time Series ML Toolkits - A machine learning toolkit integrating with scikit-learn API for time series pipelining and validation.
• Anomaly Detection - Identifies unusual data points or patterns within a time series that deviate from the expected norm.
• Standardized Interfaces - Standardizes how different forecasting algorithms handle training and prediction to ensure interchangeable model swapping.
• Machine Learning Interfaces - Implements a consistent fit-predict API compatible with the scikit-learn ecosystem for standardized model usage.
• Model Pipelines - Composing data transformations and feature extraction steps into a single workflow to streamline predictions.
• Time Series Classification - Framework for assigning categories to temporal data samples by analyzing patterns and shapes.
• Time Series Detection - Provides a set of tools for identifying irregular data points or statistical shifts within time series.
• Anomaly Detection - Identifies unusual data points or significant shifts in the underlying properties of temporal sequences.
• Classifications - Provides a comprehensive suite of tools for assigning categories to temporal data samples.
• Time Series Transformations - Provides tools to modify the structure or properties of time series data to prepare it for machine learning.
• ML Model Compositions - Sequentially chains data transformations and model estimators into single executable objects to automate predictions.
• Temporal Shift Detection - Identifies unusual data points and significant statistical shifts in the underlying properties of a time series.
• Model Performance Benchmarking - Includes standardized assessment tools to evaluate and benchmark the speed and accuracy of time series models.
• Model Selection and Validation - Provides tools for optimizing hyperparameters and validating model predictive performance within standard ML workflows.
• ML Library Integrations - Provides mechanisms to connect specialized time series tools and machine learning libraries into consistent workflows.
• Model Ensembling - Combines predictions from multiple base learners to improve overall forecasting accuracy and robustness.
• Tabular Converters - Processes raw time series sequences into structured tabular formats compatible with standard machine learning models.
• Time Series Regression - Estimates a continuous target value based on the characteristics of a time series input.
• Time Series Segmenters - Divides continuous time series into meaningful sub-sequences based on similarity or behavioral changes.
• Changepoint Detection - Locates specific points in a time series where the underlying statistical properties of the data change.
• Integration Adapters - Uses adapter patterns to wrap external time series libraries in common interfaces for decoupled integration.
• Forecasting and Analytics - Python library for time-series machine learning.
• General Machine Learning - Unified framework for time series machine learning.
• Time Series - Unified framework for time series machine learning.
• Time Series Analysis - Unified framework for time series machine learning.
• Time Series Analysis - Comprehensive toolbox for time series learning and forecasting.

Star history