• time series
• classification
• real-time monitoring
• unsupervised
• supervised
• legacy industrial machine
• data mining

This thesis work follows my internship within the startup Zerynth, on a project concerning the real-time monitoring of consumption of industrial injection molding machinery.

After an introduction of the theoretical background of the models used and the processing carried out on the data to make them usable, the thesis proceeds with the application of the models and the consideration of the results. The supervised domain is analyzed first and then the knowledge and insights learned are extended to the unsupervised domain.

The dataset was further transformed by exploiting the analysis of the consumption time series to focus on the prediction of the class change in order to be able to create a model to be implemented on the real time monitoring.