The three-way collaborative filtering recommendation introduces the decision-making philosophy of three-way thinking into collaborative filtering, achieving success in addressing the uncertainties in recommendations. However, existing three-way collaborative filtering methods still face two challenges: firstly, the determination of individual neighborhood similarity often overlooks the commonality among subjects, leading to the neglect of low-frequency items and resulting in imbalanced recommendation; secondly, using static thresholds or offline parameter tuning strategies in dynamic recommendation scenarios, where continuous feedback is received, makes it difficult to timely adjust decision boundaries, thereby increasing the rate of false recommendations. To address these issues, this paper proposes an incremental clustering-based dynamic adaptive sequential three-way collaborative recommendation(IC-DASTCR). Specifically, this model includes two innovative modules: 1) incremental clustering is introduced into sequential three-way recommendation to characterize group commonality, which improves recommendation accuracy and reduces recommendation costs; 2) a dynamic threshold adaptive mechanism is designed based on the stochastic gradient descent algorithm, for online update decision boundaries and further improve the accuracy of recommendations. Experiments and comparisons with several recommendation algorithms on multiple public datasets and real-world scenarios demonstrate that the proposed IC-DASTCR achieves superior recommendation quality and efficiency.