1.4. What is Data Science?

With an understanding of computers and computer science, it’s time to turn our attention to the last part of the book title, Data Science. It can be understood as an application of computers and computer science.

Definition

Data Science is a field that focuses on the specific problems of acquiring, curating, and exploring data, via visualization and statistical analysis, to aid decision making.

Data science as a process.

All data is initially collected in some way from the real world. Acquiring data typically involves either accessing existing data sets that someone else has collected or creating new data sets yourself. Accessing data might involve retrieving data from different kinds of data repositories that have been developed by companies, government agencies, or academic researchers. As more and more data is being collected by these groups, more and more is being made publicly accessible. As just one example, take a look at Data.gov, the U.S. government’s online repository of open data sets.

You can create your own data sets by recording data yourself or gathering it from other digital resources. Collecting your own data might involve placing sensors to monitor environmental conditions over time or regularly logging quantifiable information about your day via an app. If gathering data from digital sources, data science provides many tools that can automate collection from information sources on the internet. For example, we could generate a data set about jobs in a particular area by collecting information from job posting websites.

Curating data typically involves cleaning and organizing data sets in preparation for analysis and visualization. Real-world data sets are often messy, with missing data, erroneous information, and duplicate entries. Data science provides methods and tools for finding and correcting these problems. If data is being integrated from multiple separate datasets, data science also provides ways to properly merge different sources of data.

The large data sets we might collect are difficult for people to understand in their raw form. Visualizations and statistics are two tools that help us understand large data sets. Visualizations refer to graphical representations of data, like charts or figures. Statistics are formal mathematical tools for helping us describe and make inferences from large sets of data. Data science provides tools that allow for the automation of the creation of visualizations and statistics.

Data scientists typically apply programming along with mathematics and applied statistics to problems within a specific domain area (for example, finance or meterology). Knowing how and when to use data science tools requires expertise in the domain area where the tools are being used.

It is important to remember that data science tools only provide useful insights if the right data and the right tools are being used to solve the problem. It is easy to produce results that look important or insightful but might actually be misleading.

1.4.1. What Skills Do I Need to Do Data Science?

You will need to know programming, you will need to know statistics, and you will need to know about your domain area. (As we’ve mentioned, this book will focus on the fundamentals of the first of those: programming.) Data scientists do differ quite a bit in how much they know about or use knoweldge in these three areas. Some are stronger in programming and statistics but not domain knowledge, others may be strong in programming and domain knowledge but might not know about or need much statistics. However, one thing that differentiates a data scientist from a traditional programmer, a traditional statistician, and a traditional business analyst is the degree to which the data scientist can use tools from each of these areas to solve problems.

Another common skill that is required for data scientists is communication. Great insights and great solutions to problems are only really great if you can convince other people that they are correct. Conveying technical information to an audience that does not share your technical expertise can be a great challenge.