Jun 20 2019
This is the translation of article recently published by Asenta’s Partner-Director Luis Mauleón in Empresa XXI, summarizing the key points of my lecture last month at the Faculty of Engineering of Deusto University in Bilbao, as part of a series of events organized by Asenta Management Consultants.
Practical lessons from Michel Baudin
In an environment with an overabundance of data, it is a paradox that we still have difficulties to finding simple answers to simple questions.
The data cleaning time sink
Understanding that data is at the service of improvement in decision-making and that the volume of data generated and collected has grown exponentially, we can conclude that a significant percentage of time should be used in the cleaning and validating data. Data scientists do indeed spend much of their time preparing or conditioning data.
In turn, the increase in data volume leads to changes in analysis methods and gives a growing importance to the tools for communicating the results of observations and analysis.
Make it useful
In Manufacturing, the main focus is to use data is to better understand customer demand, improve the design of operations, speed up the diagnosis of process problems, and measure performance.
Direct observation and information systems
About data sources, Michel Baudin points out that, although the most valuable information may come from direct observation, it must be complemented with the analysis of data that comes from the information systems. About this, he highlights two frequent problems in companies:
- Manufacturing software systems like ERP, CMMS, or MES provide information for the management – planning and control – of the processes, without displaying the necessary engineering information or technical data.
- The interpretation of the data is often subjective, and sophisticated tools are useless in the absence of objectivity.
Challenges in making data talk
The five challenges identified in the use of the data relate to accessibility, quality, the analysis skills, communication methods, and integration with the improvement of the factory.
Why are they challenges? The message regarding the availability of the data is that given the amount of data, we must decide to collect only what is necessary.
The quality of the data refers to their meaning and their credibility, which will be different depending on their origin. The quality of the data from internal sources is easier to control and the information easier to understand.
Mastery of tools
For effective data analysis, we must learn a growing number of tools but also how to apply them to find answers.
The fourth challenge consists in making the manufacturing data talk. We must use visualization tools that provide relevant information in a comprehensible and agile way, avoiding “chart junk.” Some ways of presenting data are very complex, yet do not make patterns stand out or hep understanding what happened.
Finally, we must also highlight the importance in Manufacturing of visualizing the relationships between dependent variables in the processes and connecting the results with the improvement actions in the factory.
The Master Class concluded with a recommendation to companies: data scientists learn must about manufacturing and that manufacturing professionals must learn about data science.