What Is Data Science and Why Is It Such a Big Deal?
By Spencer Karp and Sameer Tirumala
What do you think of or see when you look at this data set?
Probably just a mass of numbers and strings on your screen. The picture becomes even more unclear when you find out that there are about 70 more columns and 1400 total rows in this data set. Unless you can read data like a computer, you won’t get much out of it. No patterns, no real-world application, none of it. But there’s so much information we can get out of this. So what can we do?
This is where analytics comes in. Data analytics will employ programs to visualize, aggregate, or transform data to glean descriptive insights that the human eye simply cannot.
However, most data science solutions take this one step further. Rather than looking backward and mainly providing descriptive value, data science allows decision makers to predict future scenarios and respond accordingly.
Now how could we make these predictions? We could build out a statistical model and use a function with multiple inputs to make predictions about new data. We could also prepare visualizations (like charts and graphics) for ourselves to analyze and make predictions based on what we see.
Is it that simple? Well, not really.
You may ask why we aren’t done with analyzing data after explicitly telling the computer how to interpret it. This is where the idea of machine learning, or computers discovering new information on their own, is what’s so exciting about the field of data science.
In order to model our world, we use statistical functions with inputs that represent columns of data (also known as features). Then, the way that the computer “learns” is by optimizing the function to represent how the inputs are best combined to form the output. This is done by what is known as “training” a computer. We show the computer lots of training data, and it modifies the function to fit all of the data it sees. Next, to test its fit, we give it a new challenge with “testing” data that it has never seen before. This process of withholding some data to test the computer is meant to replicate how the model will work on real-world data that we don’t yet have an answer for. After optimizing its performance, now the computer is ready to make predictions about new data based on what it has learned!
With our new understanding of how computers learn, we can return to the data set at the top of the page about real estate properties. From this data, we could input certain pieces of information (such as Lot Area, Zoning Code, or Utilities Available) and receive a prediction for what the house should be worth.
If this prediction is lower than the current market price, it could inform someone in the house search to hold off, while a bank might work harder to close a deal given this information.
But why should you care about data science? Because it’s everywhere!
In virtually every industry, firms can find applications for data science. In finance, a classifier model could predict whether a loan applicant will default or not, allowing banks to lend more safely. For retailers, a market basket model could help understand which products are frequently bought together and recommend them to spend more on related goods.
In addition, we’re now able to process and analyze new types of data beyond simple numbers in a spreadsheet. Think about images. At first, only humans could recognize images, but we’ve taught computers how to see them too. Imagine what we can do if we taught computers to drive safely on the highway. The social and economic impact of successful autonomous driving would be insane!
Data scientists use tools like statistical methods, data visualization, and machine learning models to uncover insights and generate predictions from data.
If you are looking for an interdisciplinary field where you apply analytical problem-solving in creative, outside-the-box ways, then data science is for you. Data science is already transforming the way in which we live, learn and think. Take your favorite field and think about how these techniques can be applied to it. Maybe you’ll come up with the next big idea!
