Reflecting On The Model
A little over a month and ten articles ago, I introduced a simplified model of analytics. It has been very well received. But is it any good?
In hindsight, I made the only claims that one should about a model. I have developed it for decades and I credit it for a career that tallies hundreds of millions of dollars in success. You see models can never truly be validated… only refuted. They either test well or they should be dismissed. A wise man once told me:
I am not sure about the methodologies behind your models, but the outcomes and value are evident. In the end, that is all that really matters when it comes to models— has the business benefited?
First and foremost, models are for learning. Even bad ones can grant you that, though you should be careful. Great models predict and prescribe. All models must be judged on the outcomes of those two things. The learning you think you created must be valued accordingly. It is a bit recursive… isn’t it always.
This also means that when it comes to models — there are no absolutes. There never are… But models are never 100% right or wrong, although I have seen a few that challenged the latter. It follows then that strict adherence to a model is not always required or optimal — unless you are testing it’s value and validity. I will always preach discipline, but adhering to the absolute of any model actually inhibits learning.
It also follows that no model is ever complete. It takes more than one, most of the time. And the demon, as you might expect, can be in how two or more models connect (or fail to). Alternatively, an analyst can choose to “loosen” their model. That is not a very exact phrase, but by simplifying, broadening, and allowing for some adjustment a model may become more applicable to a wider range of challenges. I call this a framework.
Frameworks may be a little less exact. They don’t always tell you exactly what, exactly when, and nothing ever tells you exactly why .They do tell you HOW. Just not exactly… more like mostly.
As a framework and a how — my model for The Analytic Process (TAP), is second to none. I have the all the outcomes I need to rest assured. But I recognize that you don’t. You are going to have to test it yourself. You can do this exactly and directly, though few have the time. OR, you can extend me your trust. If you’ve read this far, you already have. But I recognize that if you stop learning or generate a bad outcome, that trust will disappear. That is how this works!
So, providing a little foresight, the next few articles are going to dig a bit deeper into this model. Recursively, of course.
TAP #12 will examine the process embedded in all information. Coming soon…
TAP #13 will examine the organization embedded in all processes. Coming soon…
TAP #14 will build on those two anecdotal concepts to demonstrate the matrix behind the model. It may require more than one article to do that. Stay tuned to see what develops…
Gurupriyan is a Software Engineer and a technology enthusiast, he’s been working on the field for the last 6 years. Currently focusing on mobile app development and IoT.