The ice cream project: a recipe for success
Our journey began with a collaboration between Valcon, a machine manufacturer for ice cream production and one of their key clients – an ice cream producer. The goal was simple yet ambitious: achieve enhanced Overall Equipment Effectiveness (OEE) through machine learning.
But what is machine learning, and how did we use it to enhance efficiency? Think of it as crafting the perfect ice cream recipe. Just like in ice cream making, machine learning involves essential steps:
- Gather ingredients (collect data): The first step is collecting data, much like gathering plausible ingredients for ice cream
- Choose ingredients (variable identification): Next, we carefully select the right variables to create the end product, just as a master ice cream maker selects the finest ingredients
- Formulate recipe (feature engineering): With the right variables in place, data preparation begins—a process similar to crafting the perfect ice cream recipe
- Iterate recipe (model training and productionisation): Just as a chef fine-tunes an ice cream recipe through iterations and adjustments, ML models undergo refinement after being put to the test
- Produce ice cream (model deployment): Finally, you have a well-crafted model or recipe ready to be deployed to start the production of a delightful scoop of ice cream.
As simple as it sounds, the process can be time-consuming due to the abstract nature of the problem. Often, you are searching for ingredients among a pile of not-so-edible rocks or trying to use a cow not realising that it is milk you need instead. This introduces additional complexity, and you may encounter unexpected results along the way or experience unexpected “factory alarms”.
Our approach to creating the ultimate ice cream production
In pursuit of the perfect recipe (machine learning model), our team started by collecting data and creating interactive data visualisations as a prototype, highlighting essential variables like power consumption, production speed and yield. These visuals gave the client a taste of effective production analysis and issue identification and paved the way for real-time monitoring and swift root-cause problem resolution.
We employed clustering to identify various worker behaviours that affect production performance. By analysing cluster statistics, we uncovered efficiency improvements, providing valuable recommendations for how to enhance ice cream production. Clustering is a versatile tool that can be applied not only for personalising customer experiences but also for gaining insights into the variables that affect production KPIs.
Through outstanding collaboration and strong supplier relationships, the project delivered remarkable results, including a substantial OEE increase.
Let’s craft success together
Our journey through this ice cream-inspired machine learning project leaves us with valuable lessons: the importance of data accessibility, improving overall stakeholder data literacy level, and language barriers that can be resolved through data visualisations. Yet, above all, it is about savouring the experience and enjoying the journey.
With these lessons in mind, we stand ready to guide you in your future digitalisation projects. Start your digital journey with quick, accessible achievements that inspire and energise your organisation. At Valcon, we’ve proven that by blending the precision of machine learning with the sweetness of innovation, you can truly transform your business, one scoop at a time.
+45 2194 1588
Ivar Cashin Eriksson
+45 4412 1806
Anders Worsøe Gantzhorn
+45 2022 5337