Back to articles list September 4, 2019 - 6 minutes read Why Every Organization Needs a Data Analyst Kateryna Koidan Kateryna is a data science writer from Kyiv, Ukraine. She worked for BNP Paribas, the leading European banking group, as an internal auditor for more than 6 years. More recently, she decided to pursue only the favorite part of her job—data analysis. Now she is continuing her self-education with deep-learning courses, enjoys coding for data analysis and visualization projects, and writes on the topics of data science and artificial intelligence. Kateryna is also a proud mother of two lovely toddlers, who make her life full of fun. Tags: jobs and career IT job market career in data science There is so much hype around the data scientist role these days that when a company needs a specialist to get some insights from data, the first idea is to look for a data scientist. But is it really the best option? Let's see how the roles of data scientists and data analysts differ and why you may want to hire an analyst before any other role. Data Scientist or Data Analyst? So, what's the difference between data scientists and data analysts? The definitions of these roles can vary but it's usually believed that a data scientist combines three key roles—data analyst, statistician, and machine learning engineer. In other words, the companies expect data scientists to be proficient in preliminary data analysis, good at revealing causal relationships, and brilliant at building machine learning models. With such expectations, it's clear that 3-in-1 is better than 1-in-1, and data scientists become more desired by companies. But wait... is it possible for somebody to be perfect in all three roles? Even more, do all these roles require similar skills? Or are the skills and approaches used by, let's say, data analysts and machine learning engineers totally different? In fact, the approaches of these specialists do differ a lot. Cassie Kozyrkov, Chief Decision Scientist at Google, provides a brilliant explanation of this difference. She claims that data analysts are in the company to provide quick results, for example, answering the question are there any interesting correlations in the data? To meet the expectations of the decision-makers about quick and short answers, data analysts use a corresponding coding style—using fewer lines of code and producing an easy-to-interpret correlation matrix for the manager. Machine learning engineers have a totally different coding style—their goal is to build a "perfect" model, and this usually takes lots, and lots, and lots of time. Statisticians also cannot provide quick results—they'll say: "Wait, wait! We cannot draw any causal relationships from this data. We don't even know if the results are statistically significant!" Yes, sometimes you'll need statisticians or data scientists that are good at statistics to answer this kind of question. But do you really need these answers just to get an idea about correlations in the data? Actually, no. After getting preliminary results from a data analyst, you need to engage a domain expert, who decides which of the identified patterns are indeed important for business and worth further investigation. So, as you can see, it's better to have an opinion from a domain expert after preliminary data analysis and before in-depth hypothesis testing—something that is hard to arrange when the same person performs data analysis and in-depth testing of revealed patterns. You can probably already see that data analysts are in some cases even more desired than data scientists. But let's now clarify what kind of skills a data analyst should have to satisfy the needs of decision-makers and become an asset for an organization. What Kind of Data Analysts Do Organizations Really Need? The main role of data analysts in the organization is to help decision-makers by identifying interesting and important patterns in the data and by providing quick answers buried in tons of tables, graphs, and log files. In a nutshell, data analysts identify areas where you may need the attention of statisticians and machine learning engineers if the domain expert finds these areas important. So, here are the qualities you want to see in a data analyst: Data storytelling. A good data analyst can read data and tell exciting stories around the data. But what's really important is that high-profile data experts never go beyond data and always allow for a multitude of possible interpretations. For example, they can say: "It looks like we got more leads after we introduced our last advertising campaign on Facebook. This may be a signal about the effectiveness of this campaign, but the growth in the number of leads could also be due to seasonal variations. More in-depth analysis is required." Data visualization skills. The ability to create visually appealing, meaningful, and easy-to-interpret graphs is also very important for data analysts. The story always benefits from great visualizations, which make the life of a decision-maker much easier. Technical expertise. A professional data analyst can provide you with lots of interesting insights that are hidden in your data, using nothing more than spreadsheets. However, in order to provide really fast results and create professional visualizations, data analysts will usually need technical expertise beyond spreadsheets. So, today you can expect data analysts to be familiar with the programming language Python and to be proficient with such tools as Tableau or Microsoft Power BI. Coding style optimized for speed. You don't need a data analyst to have the same programming skills as software engineers or machine learning engineers. Data analysts should know how to clean data using Python, how to perform data analysis, and how to present information with clear visualizations and tables. There are some very good courses available online that teach exactly these skills. In addition, data analysts should be familiar with the most popular packages created for data analysis and use these packages to do all the analytics in the most efficient way. Learn Python for Data Science (optional). Not all companies require domain expertise from data analysts but it's definitely a skill that can be a Learn Python for Data Science of professional data analysts. In other words, if someone wants to be in the cohort of the best data analysts, they should familiarize themselves with a domain. This skill will help them to differentiate between patterns that are really important for business and findings that aren't worth the time of data experts and decision-makers. Now that we know what kind of data analysts can become a valuable asset for a company, let's summarize what we can expect from a good data analyst and why every organization needs such specialists. Every Company Needs a Data Analyst If managers of an organization make data-driven decisions, this organization definitely needs a data analyst. The companies that are lucky enough to find a good specialist with the above mentioned skills will have an expert for: collecting the right kind of data; cleaning the data; performing data analysis; presenting information with nice and meaningful visualizations; discovering interesting patterns in data, and providing insights that may require further attention from statisticians and machine learning engineers; when performing analysis, prioritizing areas that are more important for the business. To sum up, a good data analyst is a primary assistant to decision-makers that translates data into meaningful stories, provides quick answers to difficult questions, and drives business in the right direction. Tags: jobs and career IT job market career in data science You may also like Subscribe to our newsletter Join our weekly newsletter to be notified about the latest posts.