This study aims to establish a computational method to predict the processes with almost cyclostationary structure. The primary idea is relied on the estimating of the support of spectra and using the discrete Fourier transform and periodogram of almost cyclostationary processes. The simulation study and a real dataset are executed to test the performance of our proposed method. Our obtained results in this paper indicate that the proposed approach performs efficiently for both the simulated data and real dataset.