Demand forecasting powered by machine learning models for Maspex Group
- Client:
- Grupa Maspex Sp. z o.o.
- Duration of cooperation:
- August 2023 – present
- Industry:
- FMCG
Maspex Group is the largest privately owned company in the food industry in Poland and one of the largest in Central and Eastern Europe.
It is the undisputed leader in the juices, nectars, and beverages market in Poland, the Czech Republic, Slovakia, and Romania, as well as a leading producer in Hungary, Bulgaria, Lithuania, and Latvia. It is also a major player in the bottled water market in Romania. In Poland, the company leads in the pasta, jam, and sauces segments and is a top producer of ketchup, ready meals, and vegetable preserves. Maspex is also a leading producer of instant products in Central and Eastern Europe and holds the leading position in the vodka market in Poland, while being the country’s largest importer of foreign spirits.
Maspex owns well-known brands such as Tymbark, Kubuś, Tiger, Lubella, Łowicz, Kotlin, Krakus, DecoMorreno, Plusssz, Żubrówka, and Soplica. Its products are manufactured in 18 production facilities with more than 160 production lines located in countries including Poland, Bulgaria, Hungary, Lithuania, and Latvia. Over the past 30 years, the Group has built a strong market position, becoming a leader in the European FMCG sector and a symbol of high quality and innovative business practices.
Implementation objectives
Due to the impressive scale of its operations and extensive product portfolio, the company was seeking a solution that would automate the demand forecasting process while ensuring high-quality forecast outputs.
It was essential to develop advanced forecasting models that take into account different planning horizons and varying levels of granularity.
Working with Occubee and the benefits achieved
As part of the first stage of cooperation, Maspex is using Occubee’s forecasts for selected product groups.
The forecasting is based on three proprietary machine learning models, specifically designed to reflect the unique characteristics of the manufacturer.
We use Occubee’s forecasts to predict demand for our products. The system generates forecasts based on machine learning models that account for different levels of aggregation and planning horizons. These forecasts support our demand planning experts in making more informed decisions.
– says Grzegorz Woźniak, Production Planning Director, Grupa Maspex Sp. z o.o.
The first ML model generates weekly forecasts per product, covering a 13-week planning horizon. The second model generates monthly forecasts per product, with a 12-month horizon. The third ML model addresses long-term planning needs, supporting strategic forecasting across extended timeframes.
Machine learning models enable effective identification and extrapolation of sales trends, meaning that detected patterns are used in forecasting to accurately reflect changes in demand for specific products. The models also take into account factors such as seasonality and planned sell-in promotions.
At this stage, Occubee is generating forecasts for around 400 Maspex products. We deliver them in a format tailored to the requirements of the client’s ERP system. We plan to further expand our cooperation and roll out the system to additional product groups.
– says Michał Koziara, Chief Operating Officer at Occubee S.A.
