Comparison between the absolute deviation method and the ordinary least squares method for estimating the geometric regression equation

Abstract

The variation in the quality of the estimated regression models and the invalidity of using some of the models because they do not have the characteristics of good estimators leads to a lack of confidence in their predictive or estimating accuracy, which necessitated studying the geometric distribution to which the data is subject and estimating the distribution regression equation using some parametric estimation methods (the method of least squares method of least absolute deviation) for data represented by the numbers of annual secondary students in the holy province of Nineveh by 17 observations, and after comparison between the results of estimating the two methods using comparison criteria (AIC, BIC, MSE) to determine the optimal method for estimating the geometric distribution equation, the method of least squares appeared to be the best because it It has the smallest values ​​of differentiation criteria and was used for the purpose of predicting the number of secondary students for the time period (2023-2027).