"Damate" completed a project on mathematical modeling of sales

2022: Finalizing IT Solution for Sales Forecasting and Order Balancing

Damate on July 6, 2022 announced the completion of the development and implementation of an IT solution for predicting sales and balancing orders based on mathematics and machine learning. The project cost is 35.5 million rubles. Damate expects to receive additional profit in the amount of up to 61 million rubles a year only through more accurate order planning.

The project was implemented by a specially created mathematical laboratory (MathLab), which brought together specialists with mathematical education and experience in optimizing production and data analysis.

The automated production planning system includes 2 units. The first is a forecasting unit, the purpose of which is to build an optimal forecast of demand and production plan for the short term, taking into account limited production capacity and live weight. The mathematical model for building a high-precision demand forecast includes algorithms for analyzing many factors, including: order history, plans for new products and customers, current orders and promotions.

The second is a balancing unit for the selection of optimal cutting percentages, taking into account possible restrictions on raw materials, production and packaging capacities in each tier for maximum confirmation of customer orders/forecasts according to priorities.

Before the launch of the project, sales forecasts in Damat were made manually by specialists and did not take into account many factors. Their accuracy was not high enough, and production was often underutilized, the efficiency and quality of balancing orders did not meet the company's requirements at the current stage of development. In addition, if there are real orders, the task arose to determine priorities, which of them should be accepted and how much to satisfy. At the same time, it was necessary to compare the importance of the client, margin and profitability of orders.

In the system created, the actual predictive accuracy indicator is 85%. In the future, the accuracy of forecasting can be increased by adding external factors to the calculation model that affect the order forecast, including: analysis of secondary sales, interaction with retail network ecosystems and other parameters.

Automation of business processes, accurate forecast and balancing of orders from customers allows you to plan the production of specific products, order the required volume of packaging, optimize slaughter schedules, frosts, promotions, etc. The introduction of new solutions enables Damate to more efficiently use the resources of the plant, such as personnel and equipment, load production lines evenly, eliminate the human factor, and also build long-term relations with customers due to a higher degree of order coverage.

2021: Developing Sales Prediction and Order Balancing Solutions

On Damate June 10, 2021, the Group of Companies announced TAdviser its intention to invest more than 100 million rubles in the development of fully automated solutions for predicting sales and balancing orders based on mathematics and. machine learning The project will be implemented by a laboratory of mathematical modeling specially created in the company and (artificial intelligence Math Lab) and the Innovation Lab, which brought together specialists with education and experience in mathematics, production optimization and. analysis data As a result of the launch of the project, Damate expects to save at least 135 million rubles annually.

Turkey producer Damate will invest 100 million rubles in mathematical modeling
Photo: business-vector.info

The project is supposed to be implemented in several stages within the framework of the turkey production and processing direction in the Penza region. The first stage will be completed in the summer of 2021 and will become the platform on the basis of which Damate plans to further develop these solutions.

According to representatives of Damate, sales forecasts in the company are made manually and do not take into account many factors, so their accuracy is not high enough, and production is often underutilized. On the other hand, if there are real orders, the task is to determine which ones should be accepted and to what extent they should be satisfied. In this case, you should compare the importance of the customer, margin and profitability of orders. Such balancing is done manually by an expert, but its quality does not meet the company's needs at the current stage of development, noted in Damat.

"With hundreds of orders, SKUs and customers, a person faces a difficult task with millions of combinations, for which he is obviously unable to find the optimal solution. This is a task of mathematical, namely combinatorial optimization, and it is best solved by algorithmic methods, "said Yefim Boeru, vice president of mathematical modeling and artificial intelligence at Damate. "That is why we decided to develop a technology that will help business experts make optimal decisions."

According to the company's estimates, an accurate forecast of orders from customers will allow planning the production of specific products, ordering the required volume of packaging, optimizing slaughter schedules, frosts, promotions, etc. At the same time, the introduction of new solutions will allow Damate to maximize the use of production facilities, as well as build long-term relations with customers due to a higher degree of order coverage.

Solutions developed as part of mathematical modeling projects include the collection and validation of data from various sources, the analysis and development of forecasting algorithms, and the creation of reports for the business user. At the root of these solutions is a detailed understanding of business processes, accurate and complete work with data, including recording, storage, access, transmission, as well as mathematical models and algorithms.

Math Lab plans to further develop the system and develop its own advanced algorithms based on time series analysis and neural networks to improve the quality of sales forecast, as well as expand functionality for other Damate sites.