the impact of financial factors on sales forecasting with supervised machine learning methods: a case study of food and beverage companies admitted to the Tehran Stock Exchange.

Subject Areas : Computational economics

saba Alirezaei ¹ , Khadijeh Khodabakhshi Parijan ^{2 *} , Marzieh Kalaei ³

1 - Accounting department, West Tehran Branch, Islamic Azad University, Tehran, Iran
2 -
3 - Department of Accounting, Faculty of Humanities, West Tehran Branch, Islamic Azad University, Tehran, Iran

Received: 2024-10-16 Accepted : 2025-02-25 Published : 2025-04-08

Keywords: sales forecast, financial data, regression, machine learning,

Abstract :

Extended Abstract

Purpose

This study deals with the identification of financial factors affecting future sales. It aims to identify the company's financial factors to predict future sales of the food industry in the Tehran Stock Exchange using supervised machine learning methods. The study period covers 50 seasonal periods from 1389 to 1401, with a sample of 23 companies from the food industry. The financial data used includes balance sheets, production and sales statistics, cost of goods sold, and financial ratio tables. After data cleaning and removing unrelated financial ratios, about 50 financial statement items were entered into the model. Features with a correlation higher than 70% were removed, resulting in a final model with 21 financial features. The seasonal inflation rate was included as a dependent variable to predict sales for the next two periods. Objective Sales forecasting is of particular importance as one of the most important stages of financial and strategic planning in any company, especially in competitive and dynamic industries. In the food and beverage industry, which is of utmost importance due to basic human needs and sustainable demand, the ability to accurately forecast sales can help companies optimize production, inventory management, and marketing strategies. Considering the rapid changes in consumers' tastes, price fluctuations of raw materials and increasing competition in this industry, identifying factors affecting sales forecasting is very important. Financial factors, as one of the key components in the sales forecasting process, can have significant effects on the financial and commercial performance of companies. These factors include variables such as income, expenses, profit margin, capital structure and liquidity that can directly or indirectly affect sales forecast. For example, an increase in production costs could lead to an increase in product prices and thus a decrease in demand, or changes in the capital structure could affect the company's ability to finance advertising and marketing. In the Tehran Stock Exchange, food and beverage companies, as one of the important economic competitors, are facing certain challenges and opportunities.

Methodology

In this context, economic fluctuations, exchange rate changes, as well as changes in government policies can have significant effects on the performance of these companies. Therefore, detailed analysis of financial factors and identification of their relationships with sales forecast can help the managers of these companies to make better decisions in terms of sales, production and financing strategies. This research identifies and analyzes the effect of financial factors on forecasting the sales of food and beverage companies in Tehran Stock Exchange. By examining the financial and sales data of these companies, the goal is to identify effective patterns and help managers and investors make strategic decisions. In the following, we will review the available literature, research methodology and data analysis to provide a basis for a better understanding of this issue.

 Finding

 The research employs correlation analysis through regression using linear regression, random forest regression, and vector regression models. The study trained the regression model from the first season of 2011 to mid-2019 and tested it by predicting sales for the last six seasons for 23 companies. Prediction accuracy was evaluated by comparing actual and predicted values using R-squared (R²) and Mean Absolute Error (MAE). Data preparation and model building were performed using Python on the Jupyter Notebook platform.

Conclusion

The results indicated that the random forest model outperformed other algorithms in predicting and explaining sales variance using financial data. Other algorithms, including reinforcement learning, decision trees, support vector machines, and linear regression, ranked lower in terms of R-statistics.

The random forest model demonstrated that financial data explains only 38% of the sales variance for the next two seasons, while 60% is influenced by other factors. Financial factors such as total amount, previous sales amount, profit margin, debt ratio, return on assets, interest payments, operating cash, receivables collection period, profit before profit, asset ratio, current ratio, and production overhead predict less than half of future sales variance. Thus, financial data alone is insufficient for accurate future sales forecasting. Additionally, past sales information and inflation rate showed no effect on future sales predictions. Analysis of various financial factors can play a very important role in predicting the future financial performance of companies. Factors such as total amount, previous sales amount, profit margin, debt ratio, return on assets, dividend payment, operating cash, debt collection period, profit before profit, asset ratio, current ratio and production overhead, all indicate the financial and operational status of the company. and can be used as key indicators to evaluate the sustainability and future growth of companies. The combination of these financial factors can help predict the future of companies, especially when these factors are examined simultaneously and in interaction with each other. However, it should be noted that none of these factors alone can fully predict the company's future, and a comprehensive and more detailed analysis of market conditions and the economic environment is also necessary.

References:

-Emami, M. and Sohrabi Yourtchi, B. (2015). Designing a system for predicting and analyzing the sales volume of products in different geographical areas of the country using data mining methods: A case study of Soleyook Food Industries Group. (Master’s thesis). University of Tehran, Central Library - Information Hall. Registration Number: 81415. )in persian)
-Irfan, X., Tang, X., Narayan, V., Kumar Mall, P., Srivastava, S. and Saravanan, V. (2022). Prediction of quality food sale in mart using the AI-based TOR method. Journal of Food Quality, 3,1-9. Retrieved from https:// doi.org/ 10.1155/2022/6877520/
-KohyariHaghighat , A., Feyz, D., Azar, A., Zarei, A. and Deheshti Shahrokh, Z. (2018). Designing and explaining a model for brand decline in food industry with mixed method. Journal of Business Management Perspective, 17(34),53-70. )In persian)
-Lawrence, R., Perlich, C., Rosset, S., Khabibrakhmanov, I., Mahatma, S., Weiss, S., Callahan, M., Collins, M., Ershov, A. and Kumar, S. (2010). Operations research improves sales force productivity at IBM. Interfaces, 40(1), 33-46. Retrieved from https:// doi.org/ 10.1287/inte.1090.0468/
-Lamei, P. and Heidari, A. (2014). Predicting the sales volume of a new product in the food industry using a combination of quantitative and qualitative methods (Case study: Frozen pasta of Kaleh Company). Master’s thesis,University of Tehran, Central Library - Information Hall, Registration Number: 68070. )In persian)
-Moussakhani, M. and Shahravan Mehr, A. H. (2015). Analyzing the Factors Affecting Qazvin's SME's Export Development with MADM Approach.
Journal of Development Evolution Management,
7(23), 1-9. )In persian)
-Murray, F. T. and Ringwood, J. V. (1994). Improvement of electricity consumption forecasts using temperature inputs. Simulation Practice and Theory, 2(3), 121-139.
-Omidi Gohar, E. and Darabi, R. (2015). The Relationship between Earnings Variability and Earnings Forecast Using Neural Networks in Companies Listed on Tehran Stock Exchange. Journal of Economics and Business Research, 6(11),77-92.( In persian)
-Raizada, S. and Saini, J.R. (2021). Comparative analysis of supervised machine learning techniques for sales forecasting. International Journal of Advanced Computer Science and Applications (IJACSA), 12(11),102-110.
-Rohaan, D., Topan, E. and Groothuis-Oudshoorn, C. G. M. (2022). Using supervised machine learning for B2B sales forecasting: A case study of spare parts sales forecasting at an after-sales service provider. Expert System With Applications,188(3), 1-13.
Retrieved from https://doi.org/10.1016/j.eswa.2021.115925/