The catalog of magneto-morphological classes of active regions (ARs) CrAO (https://sun.crao.ru/databases/catalog-mmc-ars) was used to study 2046 ARs of the solar cycle 23 and 1507 ARs of the solar cycle 24. All ARs, except for unipolar spots, were divided into regular (obeying the Hale's polarity law, the Joy's law and the rule of the leading sunspot dominance) and irregular ARs (all others). The results of using different processing methods for cyclic variations of ARs of both types are compared. We used the following methods: a 13-rotation moving average, a cubic spline-approximation for noisy data using crossvalidation statistical technique to estimate the smoothing parameter, smoothing by the Gaussian function kernel, and the minimizing-variance method. All smoothing results are close to the results obtained using the traditional moving-average method. At the same time, the use of cubic spline-approximation makes it possible to achieve a smoother shape of the curves, smoothing by the Gaussian kernel makes the multi-peak structure of the cycle more pronounced, especially for irregular ARs.