Modeling variables affecting investment efficiency based on the combination of artificial neural networks and the optimization algorithm of cumulative movement of particles

The purpose of this study is a model for measuring investment efficiency in companies listed on the Tehran Stock Exchange based on the combination of artificial neural networks and the cumulative particle motion optimization algorithm. For this purpose, samples consisting of 124 companies listed on the Tehran Stock Exchange during the years 2008-2019 have been studied. In order to achieve the goals of the research, first by studying the previous researches in the field of investment efficiency, 17 variables affecting investment efficiency were selected and using the combined data method, the optimal model of investment efficiency was estimated based on the cash flow resulting from investment activities. The results of the model estimation showed that 8 variables of growth opportunity, quality of financial reporting, sales growth, rate of return on assets, financial leverage, operating cash flows, dividends, operating profit on total assets have a significant relationship with the amount of cash spent on investment activities. Finally, the results of comparing the research model based on the combination of artificial neural networks and particle cumulative motion optimization algorithm with other models showed that the development of the research model reduced the neural network training error with particle cumulative motion algorithm by 0.0560. Also, with the development of the research model, by entering accounting variables, the rock curvature level increases to 0.6164 and as a result, the accuracy of the research model increases to 63.602%.