The article considers peculiarities of continuous data stream clustering with the help of ART family neural networks. A generalized model of operation of an arbitrary ART family network is proposed. A methodology for planning a practical data analysis using the proposed generalized model is formulated. Implementation of the model stages for two different ART family networks is considered in detail: ART-2a and Fuzzy ART. A general approach to solving data clustering problems using the generalized model is proposed. Methods for assessment of the effectiveness of the continuous data stream clustering results are considered. Recommendations are developed for configuring the ART networks hyperparameters depending on the nature of the analyzed data.