Today Chapter 2 Intro To Analytical Thinking 1693592118

Analytical Skills
for AI & Data Science

Building Skills for an AI-Driven Enterprise
Daniel Vaughan
CHAPTER 2
Intro to Analytical Thinking
In the last chapter, I defined analytical thinking as the ability to translate business
problems into prescriptive solutions. There is a lot to unpack from this definition,
and this will be our task in this chapter.
To really understand the power of prescriptive solutions, I will start by precisely
defining each of the three stages present in any analysis of business decisions: these
are the descriptive, predictive, and prescriptive steps we have already mentioned in
Chapter 1.
Since one crucial skill in our analytical toolbox will be formulating the right business
questions from the outset, I will provide an initial glimpse into this topic. Spoiler
alert: we only care about business questions that entail business decisions. We will
then dissect decisions into levers, consequences, and business results. The link
between levers and consequences is intermediated by causation, so I will spend quite
a bit of time talking about this topic. Finally, I will talk about the role that uncertainty
plays in business decisions. Each of these topics is tied to one skill that will be devel‐
oped throughout the book.
What Is a Lever?
In the context of this book, “levers” are synonymous with “actions”
or “decisions,” so whenever we say that “we want to pull some lever
to obtain a business outcome,” this means that we are looking for
suitable actions or decisions.
Descriptive, Predictive, and Prescriptive Questions

In Chapter 1, we saw that data maturity models usually depict a nice, smooth road
that starts at the descriptive stage, goes through the predictive plateau, and finally
13
ascends to the predictive summit. But why is this the case? Let’s start by understand‐
ing what these mean, and then we can discuss why commentators and practitioners
alike believe that this is the natural ascension of the data evolution.
In a nutshell, descriptive relates to how things are, predictive to how we believe things
will be, and prescriptive to how things ought to be. Take Tyrion Lannister’s quote in
the Game of Thrones “The Dance of Dragons” episode: “It’s easy to confuse what is
with what ought to be, especially when what is has worked out in your favor” (my
emphasis). Tyrion seems to be claiming that we have the tendency to confuse the
descriptive and prescriptive when things turn out well, in what may well be a form of
confirmation bias. Incidentally, when the outcome is negative, our tendency is to
think that this was the worst possible result and attribute our fate to some version of
Murphy’s Law.
In any case, as this discussion shows, the prescriptive stage is a place where we can
rank different options so that words like “best” or “worst” make any sense at all. It
follows that the prescriptive layer can never be inferior to the descriptive one, as in
the former we can always make the best decision.
But what about prediction? To start, its intermediate ranking is at least problematic,
since description relates to the current state and prescription to the quality of deci‐
sions, and prediction is an input to make decisions, which may or may not be optimal
or even good. The implicit assumption in all maturity models is that the quality of
decisions can be improved when we have better predictions about the underlying
uncertainty in the problem; that good predictions allow us to plan ahead and move
proactively, instead of reacting to the past with little or no room to maneuver. That
said, this really is an assumption as there’s nothing inherent about prediction that
makes it improve the outcomes for our businesses.
When Predictive Analysis Is Powerful: The Case of Cancer Detection

Let’s take an example where better prediction can make a huge difference: cancer
detection. Oncologists usually use some type of visual aid such as X-rays or the more
advanced CT scans for early detection of different pathologies. In the case of lung
cancer, an X-ray or a CT scan is a description of the patient’s current health status.
Unfortunately, visual inspection is ineffective unless the disease has already reached a
late stage, so description here, by itself, may not provide enough time for a proactive
reaction. AI has shown remarkable prowess in predicting the existence of lung cancer
from inspecting CT scans, by identifying spots that will eventually turn out to be
malignant. But prediction can only take us so far. A doctor should then recommend
the right course of action for the patient to fully recover. AI provides the predictive
muscle, but humans prescribe the treatment.
14 | Chapter 2: Intro to Analytical Thinking

Today Chapter 2 Intro To Analytical Thinking 1693592118

Uploaded by

Copyright:

Available Formats

Today Chapter 2 Intro To Analytical Thinking 1693592118

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Today Chapter 2 Intro To Analytical Thinking 1693592118

Uploaded by

Copyright:

Available Formats

Analytical Skills

for AI & Data Science

Descriptive, Predictive, and Prescriptive Questions

When Predictive Analysis Is Powerful: The Case of Cancer Detection

14 | Chapter 2: Intro to Analytical Thinking

You might also like