In this work,we study the large approximate factor model where the cross-section dimension and the time dimension are both large.We revisit the existing estimation for the number of factors from the point of view the singular values of the data.The consistence of the estimation is proved with no restriction on the dependence structures between the factors and the idiosyncratic errors.Further,we study the model with time and/or individual e(←)ects and the related assumptions and consistence of the estimation are discussed.Finally,simulation experiments are conducted to demonstrate the results.