So, you want to be a Data Scientist?

 

Podcasts, Articles, & Publications

Well, allow me to provide you with a primer into the world of data analysis. Data science is like a detective for information. I sat down recently with Dr. Steven Holoman on his path to data science. Here is his take on the industry. Most data scientists begin by establishing a strong foundation at the undergraduate level, commonly majoring in computer science, statistics, mathematics, or a related quantitative discipline. Beyond the bachelor’s degree, many professionals pursue specialized master’s degrees in data science. These graduate programs typically instruct advanced machine learning algorithms, big data frameworks, data engineering concepts, and domain-specific applications. A Ph.D. (doctoral) degree can be advantageous for specific roles, especially those centered on advanced research, but requires 4 – 6 years. Overall, if you spent 4 years at the bachelor's level, 3 years at the master's level, and 5 years at the doctoral level, you would have spent an average of 12 years on the path to data science. Then, you will need to consider work experience. So, you want to be a data scientist? It is important to think deeply about your commitment.

So, what does a data scientist do? Data scientists turn raw data—numbers, text, images, and so much more—into meaningful insights and practical solutions. Imagine you are trying to understand the behavior of online shoppers, the performance of hospital departments, or the future price of a commodity. Data science provides you with a toolkit to analyze information, draw conclusions, and support better decision-making. This multidisciplinary domain combines statistical methods, programming skills, and knowledge from the area you want to understand (known as the “domain”) to address complex problems. Data science is not just about writing code or applying equations; it also involves critical thinking, creativity, and an understanding of the real-world contexts in which data-driven decisions are made. Data science has emerged in response to the enormous amounts of data produced every second by businesses, social media platforms, sensors, and even scientific experiments. According to current literature, healthcare, finance, energy, manufacturing, and public organizations rely on data science to identify patterns and guide policy or business strategies. For example, a hospital might use data science to predict which patients need the most urgent care. At the same time, an online retailer may apply it to recommend products tailored to individual shoppers, which Amazon has used very effectively through their AWS platforms. These predictions can transform raw data into knowledge, helping institutions improve patient outcomes, reduce costs, or enhance customer experiences.

Data science work typically proceeds in iterative stages. One commonly described approach includes data acquisition and cleaning, where the analyst gathers data from various sources, such as internal company databases, public repositories, or data streams from social media or other input sensors. Once collected, the data often requires cleaning—removing errors, filling in missing information, and ensuring consistency—so they can be analyzed accurately. The second is through exploratory analysis, and after cleaning, the analysis surveys the data with summary statistics, simple graphs, and visualizations. This step helps them understand what patterns and relationships might exist, offering clues about which models or analytical techniques to try next. The next step is to build the model. In this step, the researchers incorporate statistical models to predict outcomes or identify hidden groupings in the data. There are three common approaches to this step. First is supervised learning, used when data are “labeled.” For example, if you know which customers bought a product last month, you can build a model to predict which new customers will likely do the same. The second is unsupervised learning, which is applied when no predefined labels exist. Imagine having sales records without knowing what groups of customers exist. Unsupervised learning might uncover distinct clusters of similar purchasers, revealing patterns you did not know existed. Finally, reinforcement learning is similar to training a pet with treats or scolding; reinforcement learning models learn by receiving “rewards” or “penalties” for different actions. Over time, these models improve their decision-making to achieve better results. Once a model is built, data scientists check how well it performs. Suppose it can reliably make predictions or offer insights. It may be produced and used in real-world scenarios like an app or a hospital’s decision-support system. As new information flows in, the model can be retrained and improved.

Data come in various forms. Structured data resemble well-organized tables, while semi-structured data (like JSON or XML files) have some organization but are more flexible than spreadsheets. Unstructured data, such as text documents, images, or audio recordings, do not follow a set format. Each data type requires different tools and techniques; mastering these techniques is a key part of becoming a data scientist. As data volumes grow and become more complex, advanced infrastructures are required to handle processing and storage efficiently. This is where cloud computing enters the picture. Cloud computing provides flexible access to powerful computer resources—servers, storage, and specialized software—over the internet. Instead of buying and maintaining expensive hardware on-site, you can “rent” computing power as needed. This on-demand model reduces upfront costs and makes it easier for data scientists to experiment and scale their work. For instance, if you need to process a massive dataset overnight, you can temporarily allocate hundreds of cloud-based servers and shut them down the next day. Five core features highlight why cloud computing has become so critical. The first is called on-demand self-services, where users can quickly set up new computational resources without waiting for lengthy IT approval. The second is broad network access, meaning you can work from anywhere with an internet connection. Third is resource pooling, where multiple users share the same underlying hardware, making it more cost-effective. The Fourth is called rapid elasticity, where data resources can be upscaled or downscaled quickly, depending on workloads.  The final is measured services, where users pay only for what they use, just like a utility bill.

The three main ways the cloud is provided are essential in the discussion on data science.  The first is what is called infrastructure as a service (IaaS). Users manage virtual machines, storage, and networks themselves. This is ideal for those who want complete customization. The second is called platform as a service (PaaS). The cloud provides a platform (operating systems, development tools) so you can focus on building and analyzing data rather than maintaining servers. The final is software as a service (SaaS). This is where applications are fully managed by service providers like automated machine learning tools—and are available through a browser with minimal setup required. Several tools used in data analysis, such as Apache Hadoop and Apache Spark, work well with cloud computing. Hadoop can store vast amounts of data across multiple machines, and Spark allows quick, in-memory processing of these large datasets. According to recent studies, combining these technologies helps data scientists run complex analyses efficiently, even when dealing with enormous volumes of information. Importantly, becoming proficient in data science involves learning fundamental statistics, understanding data management and storage systems, practicing data cleaning, and mastering modeling techniques. Equally important is the ability to visualize results. Compelling data visualizations—through charts, graphs, and maps—helps convey insights to stakeholders who may not have a technical background.  As data science and cloud computing continue to advance, it is crucial to keep learning about the newest techniques and tools. Reading research articles, experimenting with new technologies, and practicing with real datasets will help you stay current in this fast-paced field. Whether you aim to improve patient care, streamline business operations, or support environmental research, data science offers a powerful approach to understanding our data-rich world.