In streaming data, dimension reduction is an important factor in processing multidimensional sequence data. In this paper, a double transform algorithm (DTA) is proposed. For the online sequence data, unitary transformation and hyperbolic rotation transformation is carried out respectively in DTA，and the parameters of the hypothesis function are obtained. The error values are predicted by the Newton iterative algorithm, and the optimal prediction value is obtained until the error is less than the predefined threshold. The simulation results show that, compared with the two algorithms of OGD and RON, the DTA algorithm effectively reduces the computation time under the premise of ensuring the algorithm stability.
