The user expects a new level of experience for 5G services. One of them is high data rates that can deliver smooth high definition streaming in mobility. There have been so many efforts to increase throughput per user. The Massive Multiple Input Multiple Output (MMIMO) is one of the technology enablers in 5G. However, implementing MMIMO requires a large number of antennas and costly Radio Frequency (RF) chains as well. To reduce the large number of RF chains, the beamforming technique can be considered. This paper proposes hybrid beamforming using a Discrete Cosine Transform (DCT). We put the DCT matrix in front of RF chain. Moreover, the transmitted signals may experience correlation due to adjacent antenna spacing that can deteriorate the capacity. Hence, we propose an antenna selection to maintain the capacity. We conduct simulations over a thousand number of channel realizations. We show that the capacity increases by using DCT rather than DFT. Furthermore, the antenna selection performs well over the random selection.