What on earth?
Well, this post had to be written. I knew something about blackboxes earlier, but I never thought about them in the business context. However, thinking about business and blackboxes at the same time seemed to make quite a lot of sense (at the time). I hope I convey some of my ideas to you!
Wikipedia says that a blackbox can be viewed as inputs and outputs without having any knowledge of its internal workings. Therefore to analyse or to model something with “blackbox approach” one only looks at inputs and outputs. The modelling process constructs a predictive model by using historic data. (yes, Wikipedia 2016 :D) Modelling sounds like a job for machine learning.
Blackbox with inputs and outputs
Ok ok. So why should you care about these boxes?
What makes this blackbox-thing interesting is that traditionally companies model internally known processes. One example could be a production line where you have X bolts and Y pieces going in and Z things coming out. This kind of modelling can be (easily) done in Excel spreadsheets.
Nowadays we can do way more. We generate lots and lots of data about businesses and some of the data could be used to model blackboxes. Remember that with “blackbox approach” it is enough to have data about the inputs and outputs to model the business process. This will be the catch. By applying machine learning in a business setting we can model more processes than before. For engineers more modelling is better? An example of this kind of approach would be to predict which customers are profitable.
This is my first attempt to put this idea into words, but hopefully you got something out of it.
Anyway, some questions to take with you:
How to find the black boxes for modelling (and business cases)?
Data flows and continuous processes vs DB and batch modelling?
Integration to business processes?
The future of machine learning in business development?
Thank you and keep it up!