An important driver for increasing productivity in manufacturing small and medium-sized enterprises (SMEs) is the digitization and digitalization of production processes. The associated increase in data volume offers enormous potential for analyzing and optimizing processes. Data from a variety of devices and systems increases the need for intelligent, dynamic analysis models. However, SMEs have a low to very low degree of digitalization. This is the result of a combination of various factors, such as scarce financial and human resources for research and development activities, lack of IT expertise, and a reluctance to introduce new digital technologies and artificial intelligence. Furthermore, the production processes of processing SMEs are very individual and sometimes highly specialized, so existing AI modules cannot be adapted to the existing production structure without increased adaptation effort. As part of this doctoral project, two machine learning methods were developed for practical use in a processing SME. The aim was to identify connections between energy consumption and plastic scrap and the machine settings as well as to find optimal parameter settings to increase energy efficiency and reduce the waste rate. The focus was on the simplicity of the solution and the easy adaptability to changing production processes. It could be shown that significant increases in productivity can also be achieved with less complex AI processes, the selection of which is based on a clear definition of goals.