Data Analytics: Practical Guide to Leveraging the Power of Algorithms, Data Science, Data Mining, Statistics, Big Data, and Predictive Analysis to Improve Business, Work, and Life

Chapter 16: What is Data Science?
Data science is the deep knowledge discovery that can be obtained
through exploration and data inference. Data scientists have experience
with both mathematical and algorithmic techniques, and they use these to
turn raw information into useful insights. It’s based on analytical rigor,
evidence, and building decision-making capabilities, and it gives companies
the information needed to make intelligent, well-strategized decisions about
their operations. Data science is about learning from data, and using it to
add value to information. Data scientists can work on many different
projects, including:
Predictive analytics: Predicting things such as future demand.
Recommendation engines: Software that recommends items, such as
Amazon offering a customer similar products or Netflix offering similar
television shows.
Tactical optimization: This helps businesses improve their processes, such
as a marketing campaign.
Automated decision engines: These are things like automated fraud
detection and self-driving cars.
Nuanced learning: The goal of this field is to develop a better understanding
of consumer behavior.
All of these categories are very straightforward, but the problems
involved in them are anything but. Solving them requires a deep knowledge
of machine-learning algorithms and very good technical abilities. These are
skills that take years to develop, not days, and data scientists require a
certain skill set in order to be qualified for their jobs.

---

Skills Required for Data Science
Data science is a field that requires skills from many different
disciplines, but there are three main areas you must be competent in.

---

Mathematics
In order to get any meaning from data, a scientist needs to be able to
see it in a quantitative way. Data has textures, patterns, dimensions, and
correlations expressed in its numbers, and finding those and their meaning
requires a knowledge of mathematical techniques. Business models often
need analytic models to solve, and these come from the theories of hard
math. However, you don't only have to know how to build the models, you
also need to have a good understanding of how they work. People
commonly believe that data science only involves statistics, but it is just one
of the many important math topics that need to be understood by scientists.
Statistics itself has two different branches, called classical and Bayesian.
Most people talking about statistics are referring to the classical type, but
data scientists need to have an understanding of both in order to do their
job. They also need to know about matrix mathematics and linear algebra.
Overall, data scientists need a deep and wide understanding of math.

---

Technology and Hacking
Before we get into this, I need to clarify that the hacking we are
talking about it not the top-secret spy stuff where you hack into people’s
computers and steal classified information. The hacking we are talking
about is the creativity and ingenuity needed for building models with
learned technical skills, then finding solutions to problems.
These skills are vital because data scientists leverage technology in
order to acquire huge data sets and use complex algorithms. Basically, they
need to have a far greater knowledge than just Excel. They need to know
about SAS, SQL, and R, all of which require the ability to code. These tools
allow data scientists to look at data, organize the information within it, and
use that information to come to useful insights, which would otherwise
simply stay hidden in the vast amount of data.
Hackers are algorithmic thinkers who can take a difficult problem and
find a solution to it. It is a very important skill for data scientists, since their
job requires that they work within existing algorithmic frameworks and
create their own to help solve problems.
 

---

Business Acumen 
Data scientists are, first and foremost, strategy consultants. They are a
valuable resource to a business because they have unique ways to add value
to the company. However, this also means that they need to have the
knowledge to approach and analyze business problems in addition to
handling algorithmic problems. Their value isn’t thanks to a number, it is
thanks to the strategic thinking they provide based on that number. A data
scientist also has to be able to use data to tell a story, complete with a
problem and solution. They use insights from their data analysis to craft this
story, and it is a core skill for their job.
 

---

What does it take to be a data scientist?
Data scientists need to be intensely curious, and they also need to be
able to think critically and deeply. Data science is based in curiosity,
questions, discovery, and constant learning. True data scientists will tell you
that money has never been a motive for them - it’s all about being able to
satisfy their curiosity and use their skills to solve problems.
Turning data into meaningful information isn’t just about finding the
answer, it’s also about finding what’s hidden. Solving problems is a journey
that gives data scientists the intellectual stimulation they crave while
bringing them to a solution. They love being challenged and they are
passionate about what they do.

---

Data Science, Analytics, and Machine Learning
Analytics is a commonly-used word when it comes to the business
world, and often it is used a little loosely. Analytics describes a quantitative
method of critical thinking. It is the science of analysis, the process of
gathering information gleaned from data and using it to make informed
decisions. 
 
An analyst can do many things or cover many roles, as it is a rather
ambiguous term: it can refer to a market analyst, an operations analyst, a
financial analyst, and many more. This brings us to the question of whether
data scientists and analysts are the same thing. They’re not, but analysts are
basically training to be data scientists, and they certainly have the heart of
one. Here are some examples of how an analyst can become a data scientist:
●
    
An analyst who has mastered Excel learns how to use SQL and
R to access raw warehouse data.
●
    
An analyst who is knowledgeable about stats and able to report
on the results of an A/B test learns how to build predictive models
with cross validation and latent variable analysis.
The point here is that it takes motivation to move up from being an
analyst and become a data scientist. You’ll have to learn many new skills.
However, many organizations have been successful by providing the
necessary training and resources to their analysts.
 
When talking about data science, the term machine learning comes up
fairly often. Machine learning is the ability to train systems or algorithms to
gain information from a data set. There are many different types of machine
learning, and they range from neural nets to the regression model. However,
they all have one central goal: teach a computer to recognize a pattern.
Possible uses for machine learning include:
●
    
Predictive models that can anticipate a user’s behavior.
●
    
Classification models that can recognize and filter out spam.
●
    
Neural nets that learn how to recognize patterns.
●
    
Clustering algorithms that can mine and help find natural
similarities between customers.

---

●
    
Recommendation engines that can learn about individually-
based preferences.
As you can see, data scientists use machine learning very often. They
use it to build algorithms that automate and simplify certain parts of
problem solving, which is essential when it comes to complex, data-driven
projects.
 

---

Data Munging 
Raw data has no structure and it can get very messy, and the process
by which it is cleaned is called data munging. The cleaned data can then be
analyzed and used in machine learning algorithms. Data munging requires
excellent hacking skills and the ability to notice patterns - this helps when it
comes time to merge the raw information and transform it. Dirty data often
obscures the truth, and its results can be misleading if it isn't cleaned first. A
data scientist must have excellent data munging skills in order to make sure
they are working with accurate data.

---