Presenting an explanatory model of factors which explains the effect of the company's factors on the prediction of accepted profits in the Tehran Stock Exchange using machine learning, neural network and linear regression.

The present paper deals with the way of designing a model describing the effect of corporate and macroeconomic factors on the profit forecast of companies listed on the Tehran Stock Exchange. To this end, the capability of profit forecast in stock exchange companies was investigated using four machine learning models including: support vector machine learning model of regression (SVR), simple and deep neural network (ANN & DNN) and linear regression (LM). This research is of applied research type and it is of descriptive type (correlation) in terms of data collection method, with relative data measurement scale. For testing the research questions, accounting data between 2010-2019 were prepared and input variables for the model were calculated accordingly. In order to check the results of data analysis using regression support vector models, simple neural network, deep neural network, linear regression, first the data set is divided into two training and testing parts so that 90% of the data is used for training and 10% is reserved for testing. The results of data analysis show the deep neural network model for the experimental data set has lower RMSE and MAE values than the support vector regression model. Therefore, it can be concluded that using macro variables and intra-company variables, the model of artificial neural networks can be a good approach to forecast profit of listed companies.