In order to get a competitive advantage, the number of sensors in production is rising rapidly. Companies are using data analysis to gain new insights, speed up the production, lower the price and make products with better quality. The storage of large amounts of data has become easier and economically more accessible. This raises the demands for development of analytical algorithms that are suitable to embed into automated processes and are easy to use.
This master's thesis introduces an approach for the integration of different data science environments and algorithms into a system of web services. This concept enables fast integration and is easy to use. It is suitable for domain experts (production technologists). The end user thus doesn't need expert analytical knowledge or the knowledge of associated tools. Services can be implemented in the cloud, which means more powerful hardware and access from different devices. With faster deployment and ease of use, the process is sped up considerably. The developed services can be used directly by the user, accessed through the user interface, or integrated into the automated process. Due to this integration possibilities, the services are suitable and are primarily intended for production control.
The algorithms are implemented in programming languages Python and Matlab. Web services are developed with the tool Swagger or in the Microsoft Azure Machine Learning Studio (Azure ML) cloud environment. They run on the Microsoft Azure cloud or are implemented locally. Clients are implemented in programming languages Java, Python, Matlab, Azure ML environment and Microsoft Excel spreadsheet.