A multi-input, multi-output model on fuzzy rule-based systems to predict macroeconomic variables in Iran

Abstract

The prediction of economic variables is one of the main issues in a country's macro decisions. Since in many cases there is no historical data for this purpose and it is necessary to have more than one output, it is necessary to use expert opinions and consequently, model expert opinions in the form of mathematical functions, adds to the complexity of the task and the importance of the problem. To solve such problems, this paper presents a ten-step process using fuzzy rule-based systems. At the first step, the three inputs that include: the price of OPEC oil, the level of Iran and Saudi relations and the level of political tension in OPEC member countries and also the three output variables that include: the amount of employment, the economic growth, and the oil price forecast, have been modeled in the form of trapezoidal and triangular functions. Then, these variables have been converted to linear functions. In the next steps, the three-dimensional decision tables were designed and then by using the fuzzy rule-based systems (if, ... Then...), the preconditions and sequences (results) of the decision rules were written and coded in the Matlab software. The results indicate that the outputs are in line with the existing economic realities of Iran and that three input variables to a certain extent can cause changes in the three output variables. Less technical so far with problems with this complexity of problems are capable of results with this obvious.