In the present work I have used an approach for devising provably good policies for stochastic inventory control models with time dependent and correlated demand. These models are known to be hard, in the sense that computing optimal policies is usually intractable. In turn the approach leads to policies that are simple computationally and conceptually as well as provides constant performance guarantees on the worst case expected behavior of these policies. Some other practice are also considered for the problems.