ARMA Detection of stray values in mixed models

Abstract

The analysis of time series largely depends on the accuracy of the data and does not
contain stray values that contaminate the studied time series, and affect the characteristics
of the model's capabilities, as well as its effect on the characteristics of the probability
distribution of the time series due to the deviation that occurs in the general pattern of
observations, which in turn affects the Kurtosis and skewness coefficients, which
adversely affect the model's capabilities and future forecasting power.
As for the application, the research deals with analyzing the daily global copper price
chain using the studied model (ARMA) and using the test of the proportion of places to
detect and estimate the electrolytes and then determine their type, and we have found that
the appropriate model for the studied data was AR (1) with five stray values revealed Its
type and appreciation.