Introduction: We present here a new algorithm to classify single particle trajectoties into confined and unconfined. The algorithm uses live-cell 3D single-molecule tracking of NuRD complex at any temporal regimes, such as 20 ms and 500 ms. Method: The algorithm relies on machine learning method (a Gaussian mixture model) to segment the single molecule trajectories into different classes by studying their behaviour over a sliding window of several consecutive images. The algorthim was designed to analyse the nucleosome remodelling and deacetylase (NuRD) complex, a highly conserved 1 MDa multi-subunit protein complex which binds to all active enhancers. Results: The algorithm allows to estimate from each sub-trajectory the 1-apparent diffusion coefficient, 2-but also the anomalous exponent a, 3-the localisation length Lc, and 4-the drift magnitude V (fig.1a). The anomalous exponent a (from the mean squared displacement), is particularly informative. Reference: The algorithm is part of the publication,https://www.biorxiv.org/content/10.1101/2020.04.03.003178v2 Live-cell 3D single-molecule tracking reveals how NuRD modulates enhancer dynamics by S Basu et al. This article is with minor revisions in Nature structural and molecular biology 2022. The final reference will be added once the publication is accepted.