BookmarkSubscribe
Choose Language Hide Translation Bar
elie_maricau
Community Trekker

Discovering hidden relationships in production data (EU2018 113)

Level: Intermediate
Elie Maricau, Senior Data Scientist, BASF Antwerpen

This presentation discusses a case study in a chemical production plant where the objective is to improve the product yield. As part of the project, five years of historical data (including lab data, sensor data and logbook text data) have been analyzed. The presentation tells the story of the journey we did to get to the final result and shows what critical success factors and JMP tricks we learned along the way. We found that, although JMP has some very powerful modeling algorithms build in, a data crunching approach does not always lead to success. However, pairing the aggregation, calculation, visualization and modeling capabilities of JMP with a good project workflow and input from subject matter experts has proven to be very valuable. The presentation focusses on the data analysis but starts from a clear objective and ends with concrete actions to improve the yield.

Comments
frankderuyck

Hi Eli, nice presentation, pity I was not there.. How did you specify the "stable" production days; guess no special cause signals in 24h subgoup Yield XR-chart? Is there video available of your presentation?

Article Labels
Article Tags
Contributors