It is very likely that you have been hearing about these three data specialty roles everywhere in the past few years. At first glance it may feel like they are pretty much in the same ballpark. However, in spite of some overlapping tasks among them, core responsibilities of these roles are completely different. Below you can find a very simple comparison in between each other with example cases.
Please bear in mind that shifted responsibilities between these roles or different roles may exist in different organizations which can also run quite well. A common model is selected just to be able to express basic concepts to the unfamiliar.
Data Analyst
Data Analyst is the key role for feeding information to decision makers, so that they can make data-driven decisions. They commonly work under the same business organization they support and they usually analyze data for a particular information that the business needs like:
Which customers should we target for car insurance products?
What is the cost of keeping our employees purchasing power as the same level as last year?
Presenting or reporting obtained results in the most efficient way is also a crucial part of this job. Nevertheless, they are rarely involved in the decision making process.
Skills
There is no one particular college degree required to be a data analyst. An analytical problem solver with some basic statistics knowledge can make a perfect data analyst. Data analysts should also be detail oriented, well-organized people as they need to manage and process demands from business efficiently. This makes them pretty much the project manager of their works as well. That’s why they should also have the ability to plan, prioritize and work under deadlines. Communication skills are also very useful for Data Analysts since they need to express their findings as clear as possible. A data analyst must have the following technical competencies:
Excellent command of MS Excel, MS Access
Good command of SQL (Structured Query Language)
Basic knowledge of descriptive statistics
Responsibilities
Collect raw data from multiple sources and organize them
Assure the quality of the data and doing the data cleaning when necessary
Respond to business inquiries by analyzing related data
Make use of descriptive statistics and data visualization to express retrieved information.
Example Case
Corporate Communication department of a mobile phone manufacturing company is planning to sponsor 40 different events until the end of the year. Marketing department wants to use one of the 6 featuring device models for each one of these events as the sponsoring brand. They ask for their fellow Data Analyst, Deniz to help them making this decision.
Deniz retrieves the product and the client demographics of buyers of these 6 models.
She also asks for estimated attendee demographics for each one of the 47 events.
She starts data cleaning first to standardize the demographics information she collected.
She runs a supervised clustering algorithm to group events under each one of these devices by comparing how similar customer demographics are.
She finalizes the list of recommended events for each device model, and presents her approach and her findings to Corporate Communications management.
"Data are just summaries of thousands of stories – tell a few of those stories to help make the data meaningful."
Data Scientist
The hype for this role has been enormous since 2008, that is when the term Data Scientist were first–ever recorded in literature. Popularity has been boosted further in 2011 when HBR named Data Scientist as the “Sexiest Job of the 21st Century”. Because of the long learning curve of the profession over time, and the increasing demand for data scientists created a massive income bubble for data scientists in recent years.
It would be quite accurate if we say that Data Scientists are simply self-initiative Data Analysts with advanced analytical and programming skills, business understanding and a natural-born curiosity for the discoveries. This makes them full stack data-driven problem solvers who can identify what information is valuable for the business and extract that information. They are rarely part of the business team they support and usually work under a special unit that provides services to all organization.
Skills
A degree in a related field or an equivalent training is usually required to be a Data Scientist. A business-minded computer scientist or a statistician would make a perfect Data Scientist when equipped with the necessary technical skills. Thanks to this high demand, today “Data Science” degree programs are also offered in many universities across the globe. Other than the programming and statistics background; a Data Scientist must have a strong data intuition and in-depth knowledge about the business they support. They usually have several of the followin
g technical competencies:
Predictive Modeling
Machine Learning / Deep Learning
Programming (Python, R)
Command of Big Data Tools (Hadoop, Spark etc.)
Unstructured Data Handling, NoSQL DBs
Responsibilities
Identify what information can be valuable to solve business problems and analyze data to extract that valuable information
Apply necessary post process on both structured and the unstructured data collected from different sources
Search data for patterns, correlations, classifications or dependencies and build predictive models.
Use machine learning algorithms to increase success rate of their models
Make use of advanced BI tools to communicate the conclusion of an advanced analysis.
Example Case
A credit card company wants to reduce the loss resulted from fraud cases.
They assign a Data Scientist, Akira to solve the problem.
Akira collects all the information she thinks can be related to fraud. (Like transaction data, merchant data, customer data, network information of the c
lient, previous transactions of the same merchant and customer, web or mobile app behavioral analytics data of the user etc.)
She also collects past fraud data as well and she builds a model that rates the fraud riskiness of a transaction.
She runs an unsupervised clustering algorithm to group these transactions according to risk levels.
She defines 4 risk levels. High Risk, Medium Risk, Low Risk and Safe..
Since the behavioral patterns of fraudsters are subject to change in time she also implements a machine learning algorithm so that her model learns new fraud patterns as they emerge.
She develops the model that gets all fraud related data as input and returns the fraud risk level of a transaction.
She optimizes the performance of the data model as this model will need to run before every transaction.
Payment systems developers now can use it as a decision making engine by calling this model before every transaction and act according to the risk level like "aborting the transaction and locking credentials" for high risk transactions etc.
“Data scientists are kind of like the new Renaissance folks, because data science is inherently multidisciplinary.”
Good work
Okk smjh gya