In small signal analysis of amplifiers,current gain,voltage gain,input and output impedence are found.My question is" why these parameters vary in approximate model method and h-parameters method?"

 

Because every collection of components used to assemble an amplifier has a unique set of parameters. When combined they give different results. As a simple example take a voltage divider with two 1% resistors. Compute the ratio with nominal values. Then recompute with extreme values. Get the idea?

 

Because models are models, they are simplifications and they do deviate from the real things. If you take two models, even if they predict the same things, the results would not be 100% in agreements. The only question is, do the results differ wildly between models or do they agree within the approximate modelling tolerances?