For power estimation in power systems, the power flow formulation is the most commonly used method. Data-driven approach is a newly proposed technology, unlike traditional model-driven power flow calculations, it can bypass circuit parameters and environmental variables to obtain results from global bus data. However, with the development of renewable energy, more complex power grids make it difficult to obtain complete data and reduce performance. Therefore, this paper proposes a method of estimating power from incomplete bus data. The method is based on a modified data-driven power flow model, and its core is to derive and induce a regressive model through the physical model. The simulation results on IEEE standard bus 30 show that this method has higher accuracy than other models in incomplete bus data power estimation.