Pert data differs significantly from the rest of the samples, with negative effects in all areas such as classification, clustering, and regression issues. To deal with the impact of mismanagement data, it should be considered that their impact is reduced and a so-called robust method should be provided. In regression problems, there are different methods for detecting Perth data, most of which identify Perth data against these data during the regression operation for retaliation; in fact, they estimate the decision level that is not affected by Pert data.
Most input field methods take into account the correlations between inputs, but in this paper we seek to transfer data from the input domain to the frequency domain using the Fourier transform so that we can provide mathematical relationships to data that behave differently at other frequencies. Be identified.
In this case, as a preprocessor without regression, we can identify and delete Perth data. The innovative solution introduced in this paper will be tested on simulated data and time series data sets and compared with robust regression methods proposed in the knowledge frontier. will be.