Time series clustering method links clusters to actual waveforms
Seeing Time: Visual-Temporal Representation Learning for Interpretable Time Series Clustering
Machine Learning
Summary
Finding meaningful groups in complex time-based data is hard because patterns are difficult to interpret. The authors created WAVE, a method that pairs time series data with pictures of their waveforms to form clusters that can be directly traced back to recognizable wave shapes. This helps people understand and trust the groups found in the data by showing examples of actual wave patterns from each cluster. Tests on many datasets show that WAVE finds clearer and more interpretable groups than previous methods.
What this means in practice
- •For industrial monitoring teams: Group complex sensor readings by waveform patterns that can be visually inspected to identify operational states or faults.
- •For financial data analysts: Cluster multivariate stock or economic time series while tracing each group back to typical price movement patterns for better interpretability.
Authors
Zheng Zhu, Zexi Tan, Yuming Deng, Yiqun Zhang
Abstract
Multivariate Time Series (MTS) clustering is an important tool in temporal data mining, aiming to discover latent group structures from complex observations without supervision. Although existing deep clustering methods can learn discriminative temporal representations, the resulting latent clusters are often difficult to relate back to waveform characteristics that practitioners can directly inspect and compare, limiting their ability to assess whether the discovered patterns reflect meaningful temporal behaviors. This paper, therefore, proposes WAVE (Waveform Aligned Visual-temporal Embedding), which treats time series and their deterministically rendered waveform plots as complementary views of the same observations. To produce discriminative representations whose cluster structures can be traced to observable waveform characteristics, WAVE aligns and integrates fine-grained temporal variations with holistic visual patterns, while associating each discovered cluster with its centroid-nearest authentic sample. Accordingly, interpretability in this work specifically refers to waveform-level traceability rather than a general explanation of model decisions. Extensive evaluations across 10 real-world public datasets show that WAVE achieves the highest macro-averaged clustering performance and the best average rank among the compared methods, while qualitative case studies illustrate how the discovered clusters can be inspected through authentic waveform records. The source code is available at https://github.com/Zheng-Zhu1/WAVE.