Webinar dates

Learning Objectives

The students will take away strategies and tools to do the following:

Intended Audience

This webinar will be most useful to the following:

Webinar Outline

About Michel Baudin

Michel BaudinMichel Baudin's background gives him a unique perspective on manufacturing data mining. Like most contributors to the art of manufacturing, he is self-taught in this field. Before walking factory floors, however, he was a researcher with published, refereed papers in applied math, and a software engineer involved in specifying, developing and implementing manufacturing execution systems (MES). When he started in Lean in 1987, his Japanese mentor Kei Abe taught him how to analyze plant data from printed computer reports using pocket calculators and drawing charts by hand. In the 90s, this work migrated to electronic spreadsheets and then to relational databases as they became available on PCs.

Michel now has 30 years of industry experience, including 24 in consulting, training and writing about Lean. Since 1987, Michel has helped  European, American, Russian, Chinese and Japanese companies in industries ranging from automotive, aerospace, and electronics to food and personal products.

Data Mining:

Making Factory Data Talk

To run manufacturing operations, you need simple answers to simple questions.

You need to know what you are making, in what quantities, how volumes change over time, how long it takes you to fulfill orders, how much of your output is defective and why, how well you equipment and people perform, etc. You can find some, but not all the answers, by directly observing the shop floor and listening to  managers, engineers, and operators. Your eyes only see the present, and  what others tell you is just their perceptions: you must also mine the data for the nuggets of information that neither your eyes nor personal communication will give you.   

Production is driven by orders converted to schedules, it is performed according to specs, its status is monitored and its history is recorded both in terms of quantity and quality.  The challenge is that, in most factories, the data resides in multiple, archaic legacy systems and ad-hoc spreadsheets,  with inconsistent nomenclatures, input errors, and an overwhelmed IT staff convinced that nothing can be done until a new, all-encompassing system is implemented two years from now. 

But two years is an eternity. Short-term improvements in the use of existing data are necessary (1) to set and implement a strategy now, and (2) to specify requirements on new systems later. Such improvements are feasible at a fraction of the cost of a new system, by leveraging  tools that are either already present on engineers’ laptops or can be downloaded at little or no cost. 

Retrieving information out of data collected for a different purpose is data mining. Retailers or airlines do it routinely on commercial transaction data collected every day in volumes that far exceed those of manufacturing operations. As a result, while manufacturing data requires cleaning before it can be trusted, it can be made to talk with tools that are simpler and easier to use than in retail or e-commerce. Then the results must be presented effectively to lead to action. 

Few manufacturers are effective at extracting information from the data they collect. These few make better decisions on capacity planning, production line design, the allocation of work among lines and machines, or supply chain management. They anticipate the quantitative impact of these decisions before implementing them, and validate the results afterwards. With the tools presented in this webinar, you can do it too.