What is Data Mining?
Data Mining means „digging into data“ and extract information. In former days we did it to find gold but now it has been substituted by data because more and more data are generated driven by the digitalization but also by the increasing complexity of intelligent products and processes. These data contain a very big and worthful knowledge which can be useful for different purposes, for example for model building of root cause relationships.
Data Mining detects patterns in data and knowledge about influences, complex dependencies, the generation of decisions inside of complex systems with a high amount of influencing variables and even more.
How to apply Data Mining?
The standardised Data Mining process follows the CRISP (which stands for “Cross Industry Standard Process for Data Mining”) data model which is applied by iterative steps. CRISP-DM is a proven method for the construction of a Data Mining model.
1. BUSINESS UNDERSTANDING
Fixing goals and requirements, definition of the concrete project task(s) and the steps as necessary, project schedule
2. DATA UNDERSTAND
Data collection and first review of available data, explorative analysis and checking potential problems with the quality of data
3. DATA PREPARATION
Selection and cleaning of data in order to develop the final data set for modelling
4. MODELLING
Application of suitable Data Mining tools, parameter optimisation, development of different models
5. EVALUATION
Selection of the best fitting model with respect to the overall task, project review and definition of the next sense-full steps
6. DEPLOYMENT
Finalisation and presentation of the results, integration of the model into any decision making process, project review
Use Case
Uses of Data Mining for process optimization
A bigger amount of mechanical, electrical and electronic components has passed an assembly process consisting of several steps with many influencing process parameters. The output was an assembly group which has been tested by an EOL tester. In detail there are measured more than 400 Y-parameters and the final result of their status is an „ok / not ok“ decision. You can see that we have a very complex root cause system and its complexity will raise again by the dimensional values of many moulded injection parts. It’s not helpful to think about any DoE (Design of Experiment) in such a situation.
An alternative approach in order to learn about such systems could be a decision tree. It may deliver transparency.
An extended view of component criteria combined with the EOL parameters can detect potential correlations between dimensional criteria and testing parameters.
Also it is possible to find critical parameter combinations and causes for failures if all of the process parameters are part of Data Mining as well.
An additional benefit of Data Mining in such projects is to find decision paths and to use them in order to deduct limits or tolerances in complex systems. It will be very interesting to deduct tolerances for single parts that are really needed when interacting in a complex design.
Why use Data Mining with MTS Consulting Partner
- For the start of our cooperation, we recommend one-day workshop for:
- describing the existing situation regarding processes
- Key Performance indicators, problems and existing improvement approaches and initiatives
- drill down of problems and challenges, assignment of suitable approaches (like Lean, Six Sigma, DFSS, Data Mining, Process Mining)
- identification of Data Mining projects
- Definition of Data Mining projects, support in the selection of necessary analysis tools and Coaching for the whole project based on CRISP model
- Implementation of Data Mining into your organisation and definition of the interfaces
- Goal oriented connection of Data Mining with Six Sigma, Design for Six Sigma and also Process Mining if necessary
- Skills analysis – systematic deduction of your specific qualification requirements in all business segments and departments, based on roles, tasks and responsibilities
- Development of a customized Data Mining concept and selection of necessary methods in your company
- Goal oriented connection of Data Mining and your Industry 4.0 concept
- Trainings for Data Mining (basics and advanced)
- Iterative approach: analysis, implementation, and verification. We divide a complex problem into many small parts for which we study solutions.
- Prototyping: we evaluate the feasibility of a project through a limited investment, reducing the risks linked to the creation of an activity.
- Documentation: we offer the customer the overview of what we are going to implement.
If you are looking for a Data Mining consulting session, MTS Consulting Partner will be glad to offer its vast experience.
Let’s get started!
We’re looking forward to providing you with a quote, a demo or any information about our services.