A Mixture Model Approach for Clustering Time Series Data

Author(s): Shenggang Li

Originally published on Towards AI.

This member-only story is on us. Upgrade to access all of Medium.

Time Series Clustering Using Auto-Regressive Models, Moving Averages, and Nonlinear Trend Functions

Clustering time series data, like stock prices or gene expression, is often difficult. Methods like K-means with correlation distance may group data by shape but overlook more complex, evolving patterns.

For instance, stock prices don’t move randomly, and gene expression levels react to complex biological processes. Both require a more thoughtful approach.

That’s where the mixture model comes in. It helps by using AR models, MA, and trend functions to efficiently cluster time series data, revealing core relationships and patterns.

Predicting these kinds of time series is hard, since stock prices often follow a random walk, and gene expression can fluctuate for many reasons. But by clustering these curves into meaningful groups, we can uncover patterns that help with modeling or decision-making.

This method shows us what’s happening under different conditions, providing a better understanding of dynamic data. Whether in finance or biology, this approach reveals hidden structures in the data that simple clustering methods tend to overlook.

When you have a lot of time series data, like sales in a supply chain, gene expression, or stock… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI