| The complexity of modern industrial processes and the constant innovations in production monitoring technologies and data collection, strongly outline the need for advancements in production data analysis. Data Mining is a rapidly growing field, aiming in understanding data and extracting previously unknown information, with the use of Machine Learning techniques, in order to optimize production. Our cooperation with Johnson & Johnson, enabled us to obtain and explore real case production data. The exploitation of them focused on two distinct goals. The first one was the visualization of the data and the graphical representation of all variables characterizing the mixing process, for an improved data overview. The second one was the Machine Learning algorithms’ modification and application on the properly pre-processed production data, to look into the possibilities of these techniques in the enterpise space. Machine Learning, by definition, sets to represent data as objects in space, utilizing labels and distances between them. In this direction, data were grouped into objects and vectorized with various techniques. Classification and Clustering algorithms were parameterized and implemented, investigating unique attributes of the provided data. Distance calculation methods for each algorithm were examined in depth, and each experiment was assessed ”using” different evaluation metrics, in order to examine the performance of the algorithms and the result’s accuracy, compared to the initial data. Our conclusions indicate that Machine Learning can drive important business decisions, through process quantification, and further research can be done in each specific case.
|
PDF] [
BibTex] [
Print] [
Back]