GSK wanted to leverage a large archive of unused process data to better understand their manufacturing processes. This data consisted of unsegmented time-series data for a large number of production batches of toothpaste, along with test results for each batch.
In collaboration with MathWorks Consulting Services, they developed a tool based on Deep Learning Toolbox for automatically cleaning, segmenting, and interrogating this time-series data. This tool is accessible via a web app, which enables even nontechnical users to quickly visualize and compare different datasets.
GSK has used the tool to test their understanding of their manufacturing processes and to drive experiment design aimed at improving these processes.
- Ability to link different data files and formats together
- Quickly iterate code throughout development
- Consulting support throughout process
- Ability to build GUI to enable nontechnical user to leverage data