Show HN: Efficient Modeling of Nonlinear Dynamics in Nonstationary Data Streams

Category: library

Tags: koopman-operator, time-series, online-learning

Score: 6.0/10 (Innovation: 7, Technical: 7, Documentation: 6, Utility: 4)

AdaKoop implements a novel method for modeling nonlinear dynamics in nonstationary data streams using Koopman operator regression, designed for online forecasting with adaptive model selection. It is interesting because it targets a challenging real-world problem (nonstationary, streaming data) with a theoretically grounded approach from the KDD 2026 paper, though the current codebase is early-stage and lacks extensive documentation or examples beyond a minimal demo.

Target audience: data engineers, data scientists

Repository: https://github.com/C-Naoki/AdaKoop · Python

View on Hacker News