Minitab is a professional software dedicated to quality Statistics, or applied Statistics. It’s the leading tool for many types of industrial data analysis:
- Design of Experiment (DoE)
- Analysis of Variance (ANOVA)
- Capability Analysis
- Control charts
- Reliability and stability
…and really much more.
Well, all of these methods are used in order to get useful info out of data and then take decisions. Now, turning data into info, in order to improve your business, is exactly what Data Science is aimed to do…
Thus, we can reasonably state that Minitab is a good tool for data scientists.
But Data Science is much more than statistical analysis, so what could we do in order to integrate Minitab into a real Data Science workflow?
In Kiwi Data Science we are often working with Minitab and what we are doing is to project proper environments surrounding Minitab in order to acquire and validate data and to go further with simulations, machine learning and classifiers.
Data validation is a process that can be improved with many methods from Data Science; validating data is really important in order to have Minitab working on good and properly ordered data.
Minitab can create predictive models with regression methods, but we can complement Minitab with other predictive models, which work side by side with it. Prediction can help saving time and money and Data Science is so much about prediction, so why shouldn’t we improve the whole system giving Minitab some external help?
Moreover, usually companies do not work with real database management systems, they use Excel, which is the worst tool you could use to collect and share data. Using a real database in your workflow really helps! With a database, data can be properly collected, shared and validated using historical data; Minitab can query the database and work with organised and clean data in real time; also, having all the data together makes it possible to plan more complex analysis and to train predictive models, other than regressions.
I think it’s time for some kind of revolution within the world of quality control and improvement; Statistics cannot face the problem alone, neither Six Sigma can. We need Data Science to develop a brand new workflow which can add real values to the important field of process control and improvement.
These added values could be listed as follows:
- smart data validation
- prediction and classification
- data visualisation
- workflow optimisation