v0.1.0

The primary goal of AMPIIMTS is to facilitate motif discovery and anomaly detection on complex datasets like environmental sensor readings, where sampling intervals may be uneven and multiple variables interact. It provides a high-level pipeline to ingest raw time-series data (e.g. CSV files with timestamps), preprocess and normalize the data, compute the matrix profile, and return the locations of noteworthy patterns or anomalies. In essence, AMPIIMTS streamlines a sophisticated time-series analysis workflow into a few function calls, making it easier for practitioners to analyze irregular multivariate data. This document gives an overview of how the package works, its internal modules, and how the results (motifs and anomalies) can be interpreted considering real-world events.

Features

• Data Preprocessing
• Automatic Window Selection
• Clustering of Dimensions
• Matrix Profile Computation
• Motif and Discord Discovery
• Iterative Refinement (Smart Interpolation)
• Visualization Utilities

Documentation

• Documentation
• Examples with air bejin dataset
• Diagram

Diagram