Abstract This article shows a comparison of various methods as a statistical technique for making applied forecasts of Colombian coffee demand in South Korea. The aim is to model the demand behavior in the most adjusted and efficient way possible. To do this, the correlation factor between demand and different macroeconomic variables was analyzed, the one with the greatest relationship is selected and the autocorrelation factor is evaluated. Later, different deterministic methods are used such as linear regression by least squares, simple moving average, weighted moving average, simple exponential smoothing, and exponential smoothing with trend. As a result, a multiple linear regression analysis is obtained, an evaluation of the predictive capacity of the regression model was made through analysis of variance, and the calculation of the standard error of multiple estimation, multiple determination coefficient and the adjusted determination coefficient.