Observations of chromospheric activity variations for some lower main-sequence stars from the Mount Wilson Observatory's HK project reveal a cyclic behavior comparable to the sunspot cycle. Even in the relatively short interval that they have been observed, those stars show stellar cycles and other features, like grand minima. The quasi-periodic nature of such variations is not completely compatible with the standard Fourier analysis, so we applied a wavelet analysis to study the nature of regularities in the data. We computed wavelet transforms and energy spectra for the 25 yr records of surface magnetic activity in four stars: HD 3651, HD 10700, HD 10476, and HD 201091. We present a modified wavelet technique that is suitable for analysis of data with gaps and find that the common aliasing problems due to the finite length of the observations and irregularly spaced gaps between data can be reduced on both large and small scales by applying this algorithm.