Agglomerative likelihood clustering

Abstract We consider the problem of fast time-series data clustering. Building on previous work modeling, correlation-based Hamiltonian spin variables we present an updated non-expensive agglomerative likelihood clustering algorithm (ALC). The method replaces optimized genetic based approach (f-SPC) with recursive merging framework inspired by in econophysics and community detection. is tested noisy synthetic correlated datasets a built-in cluster structure to demonstrate that produces meaningful non-trivial results. apply it as large 20 000 assets argue ALC can reduce computation time costs resource usage for scale applications while being serialized, hence has no obvious parallelization requirement. be effective choice state-detection online learning non-linear environment, because requires prior information about number clusters.