The food industry has always run on very tight margins, including both the takeaway and eat-in food sectors. It is also an industry in which demand planning is very important. Hygiene regulations—and the nature of perishable foodstuffs—mean that food and ingredients must be stored correctly. However, they still cannot be kept indefinitely, even when refrigerated or frozen. Indeed, some foods can be kept for a few days at most, and many for less.
Companies therefore cannot afford to over-order ingredients, especially perishable goods. If they do so, they will end up throwing out stock and losing money. Equally, however, if they do not have enough stock and ingredients available to enable them to meet demand, they will have unhappy customers. These customers may well move permanently to competitors, and leave negative reviews, both of which have implications for profits.
Forecasting demand and delivering supply
Ultimately, food companies need systems in place to provide them with accurate forecasts of demand. These systems must be sensitive enough to predict busy times of day, and also highlight events that are associated with an increase in demand. They must also be able to predict variations in the type of demand, for example, for seasonal goods. This will enable the owners and managers of restaurants, takeaways and cafés to ensure that they can order the right ingredients, in the right quantity, and at the right time, to meet demand.
A team from SAS partner DataCurate decided to bring this problem to the 2024 SAS Hackathon. The team wanted to develop a system that would enable food businesses to decide what ingredients to procure, when to procure them, and in what quantities, to meet demand. The plan was to support these companies to increase the efficiency of their operations, and particularly to minimise waste.
The team used a two-stage approach. In the first phase, they aimed to forecast demand at store level by using historical sales and offers data. They also wanted to aggregate store-level forecasts to provide demand forecasts for central warehouses linked to larger companies. In the second phase, they wanted to move onto accurate procurement planning. This aimed to optimise procurement decisions by considering the future prices of ingredients, warehousing costs and product shelf-life. The focus overall was on timely procurement, minimising total costs and avoiding excess stock.
Using SAS to deliver results
A pizza company supplied the DataCurate team with data from 30 separate pizza restaurants or stores, covering 13 different types of pizza with 28 different ingredients. The team analysed this data, looking at store-level sales, promotions at store and company level, and seasonal trends.
The team drew on a range of SAS solutions to help them develop their procurement engine. They used SAS Data Explorer for data preparation, and SAS Visual Analytics to generate insights. They then added Model Studio for forecasting, and SAS Studio for customer optimisation algorithms. They created a model that optimised procurement decisions looking at demand forecasts, ingredient prices, shelf-life and storage costs. Finally, they used SAS Graph Builder to display their results.
As it stands, the DataCurate model will enable the pizza company that supplied the data to improve its procurement decisions about both store and warehouse level. It should facilitate more informed buying decisions, including both timing and quantities. This will enable stores within the company to avoid running out of stock, and wasting ingredients, particularly those with a short shelf-life.
More importantly, however, the model has much wider potential impact. With different data inputs, it could be used by other food companies and outlets, including both restaurants and takeaways. The combination of central warehouse and distributed stores is widely used in the industry, making this solution applicable around the world, especially for larger companies.
Creating an impact
As Royson Almeida, the DataCurate team leader commented,
“The company can ensure efficient operations and minimal waste.”
Given the tight margins in the industry, this may be all that many food companies require to give them a competitive advantage. However, a competitive advantage is not the only aspect of this solution that is important.
As the global population expands, we cannot afford to waste food. Better procurement decisions in the food industry help to ensure better use of global resources. This may only be one company at the moment—but the DataCurate solution has much wider potential impact. Accurate demand forecasting is an important first step in avoiding food waste.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.