Clustering events in a sequence

I have a sequence of recurring events that I would to group together into representing different operation activities of the underlying process.

  1. These events may have an order in their occurrence; or maybe not. Consequently, I would like to explore and investigate if any relationship exists between the events. Are there any better methods than using Hierarchical clustering?
  1. I might want to build a model that can determine the operational activity based on the events it recognized as belonging to the clusters representing them operational activities. Maybe HMM, RNN, etc.

I have been searching for an algorithm that lends itself to the problem that I am trying to solve. I came across a few that might be adapted to be able to solve the problem such as HMM, RNN, or clustering algorithms.

I am not looking for a solution, but for advice and guidance in the right direction towards solving the problem.

I appreciate any insight you may have to share with me.

Topic sequential-pattern-mining rnn dbscan time-series clustering

Category Data Science

1
answered at

If the data is unlabeled, you'll have to apply clustering. One useful way to frame the problem is as time-series clustering. Almost all clustering algorithms have a time-series version (e.g., k-means and hierarchical). The choice of clustering algorithm depends both on the type of data available and purpose of the project.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.