Primer: All You Need to Know About Data Science

Jul 30, 2020 | by Levon Paradzhanyan

Artificial intelligence (AI), together with its brethren buzzwords data science, machine learning, and deep learning have been around for some time now and are no longer future concepts. Yet misconceptions persist about the true meaning of these terms. Some think they mean the same thing, even developers struggle to understand the difference.

In part one of this series on finding the true meaning behind these buzzwords, we focus on data science – the overarching discipline that all the other methodologies fall under. Each article will explore these technologies and the one thing they all have in common – data.

Automation is nothing new

For centuries, man has invented new technologies that automate previously manual activities. While they lack human-like intelligence, there are many examples in ancient civilizations of the automaton or self-operating “intelligent” machine (today called a robot) that automatically performs a sequence of tasks.

But automation isn’t limited to robotics. Our lives are increasingly empowered by smart data-driven technologies. Netflix, Spotify, and Amazon recommend content and experiences that resonate with user preferences. Companies are also gathering human behavioral data to improve and innovate new services, detect and prevent fraud, and more.

Automation also helps us get around. Tesla and others have developed intelligent, self-driving vehicles that learn from the world around them, constantly improving autonomously.

Data is the fuel that powers all of this.

Data holds answers

The amount and complexity of data being gathered in the world today is impossible to fathom, hence the term “big data”. But big data alone doesn’t deliver business-impacting analytics and insights, it simply refers to a large amount of data sets.

Rather than quantity, it’s the quality, accuracy, and completeness of data that really matters – plus a good approach to data analysis. At EastBanc Technologies, we’ve long argued that big isn’t always better. Small sets of data can yield rapid results and it all starts with a minimal viable prediction.

Industries that derive notable benefits from data include education, media and entertainment, government, transportation, and banking. Data, information science, and business intelligence help these sectors uncover trends, make predictions, and improve decision-making.

Successful data, as we’re witnessing first-hand, can help public health organizations understand what preventative measures society can take to minimize the risk of pandemic outbreaks. In education, data can steer a student along the right career path based on academic results, strengths, and interests. Data also helps predict much of what we take for granted, from the weather to the precise location of potential traffic congestion during public events, traffic incidents, or holidays – make it tangible in the form of data visualization.

But statistical analysis of data can also hide patterns and insights that we’ve yet to think about. For example, data could reveal patterns that fraudsters use to embezzle money from an online banking system. It can also reveal likely buying habits, such as men who purchase diapers have a greater propensity to stop by the beer aisle too.

But data is a beast

Today, data is the life blood of any business. That’s why understanding data and finding new patterns, hidden correlations, and insights can impact the success of the business.

But data is complex and challenging to work with. Organizations gather data all the time, but it’s often unstructured and stored across different systems – CRM, marketing automation platforms, analytical software, log files, the cloud, and so on. Despite what data architects and data analysts will tell you, there is no universal tool for intelligently determining the relationships between these systems and processing the data within them in huge volumes.

That is the role of the data scientist.

What is a data scientist?

A data scientist is so much more than a data analyst. Data scientists juggle many fields and disciplines. They are part-detective, business analyst, and software developer. They can dive into any industry, analyze it, and develop predictive models that drive the business forward.

The data science lifecycle navigated by the data scientist looks something like this:

Let’s consider the elements of each phase:

  1. Discovery

    : During this phase, the data scientist takes time to understand and analyze the objectives of the business and any challenges.
  2. Data preparation and mining

    : Now the data scientist collects data from various sources, cleans it, and manipulates the data into a format suitable for machine learning algorithms. This phase also involves the time-consuming task of making assumptions or examining hidden patterns in the data (also known as the creation of a new feature or individual measurable property).
  3. Modeling

    : Here the data scientist trains and evaluates the performance and accuracy of machine learning models.
  4. Visualize and communicate results

    : Diagrams and other methods are used to explain findings in a clear and consumable way.

Data scientists can work on many initiatives and business use cases. They help marketing teams improve their strategy based on consumer behaviors. They can also help optimize price formation through analysis of consumer purchasing power, public trials, sales history, competitive pricing, and product popularity.

How does a data scientist differ to a machine learning engineer?

The data science lifecycle can be completed by the same person, however in many cases a machine learning (ML) engineer takes on the role of engineering the model. This is commonplace if the data scientist lacks the programming language skills needed for full-stack solution development.

Let’s break down the distinct roles of the data scientist and ML engineer:

Without machine learning, data science would not be possible.

In summary

Data science has made it possible to derive insights in ways that were not previously possible. It helps organizations overcome the challenge of unstructured and siloed data and the limitations of traditional data analytics tools. With the help of predictive and prescriptive models, businesses can unify and process data to uncover hidden patterns, improve decision-making and business outcomes.

Stay tuned for the next in this series in which we explore machine learning and how it compares with data science.