Time series clustering improved with faster better feature extraction
CLUES-WEASEL: No additional clues required to choose your time series clustering algorithm
Machine Learning
Summary
Grouping similar time-based data is hard because fast methods often give worse results, and accurate ones are slow. The paper introduces CLUES-WEASEL, a new way to turn time series data into features without needing extra hints and then cluster them efficiently. The authors show that their method works better and faster than existing approaches. They also find that their approach can work well with other feature extraction methods.
What this means in practice
- •For data science teams: Cluster diverse time series datasets more accurately and quickly without manual feature crafting or tuning.
- •For industrial monitoring teams: Group sensor-generated time series data promptly to detect patterns and anomalies in manufacturing or infrastructure.
Authors
Johann Faouzi
Abstract
Time series data is very common in many real-world applications and in numerous domains, with increasing interest for automated information extraction using machine learning. One of these subfields is time series clustering, which consists in identifying clusters among a set of time series in an unsupervised fashion. Most time series clustering algorithms suffer from the same balancing act: they trade clustering performance for faster runtimes or vice versa. We present a novel time series clustering algorithm that we call CLUES-WEASEL, which stands for CLustering with the UnsupervisEd Second version of Word ExtrAction for time SEries cLassification. CLUES-WEASEL extracts features using the unsupervised version of the transformation step of WEASEL 2.0, which is a time series classification algorithm, then reduces these features using principal component analysis, and finally performs clustering with the $k$-means algorithm using these reduced extracted features. Through extensive experiments, we prove that CLUES-WEASEL is significantly better than any other existing time series clustering algorithm while being (much) faster than any state-of-the-art one. We also show that the architecture of CLUES-WEASEL can work well with other time series feature extraction algorithms. Our findings highlight the relevance of CLUES-WEASEL for time series clustering.