A data scientist is a highly demanding job today—and it is difficult to find a ‘qualified data scientist’. When we say ‘qualified’, does it mean that a data scientist must have a solid background in coding? Or, is it possible to break into data science without that background? Although most companies look for a qualified data scientist, it does not make a non-coder less eligible. Ultimately, what a business firm needs is to get work done either through a coder or a non-coder. Let us see what some experts, developers, providers and bloggers on data science have to say.
Carly T, Senior Manager, Security Strategy & Expert Engineer, Machine Learning Activision, says no-code data science solutions are more popular lately, and there is an increase in demand for such data scientists. This demand has now diluted the title of ‘data scientist‘.
Many great enterprise data scientists began their careers in data science without any prior knowledge or experience in coding. It does not matter how much coding you know when you pursue to become a data scientist, but it is important that you understand basic programming that includes mean loops, functions, and if-else programming logic.
Besides this, there are plenty of data science courses to upgrade coding skills.
The basic requirements for a non-coder to become a data scientist include:
- Thoroughly understanding probability and statistics.
- Having a passion for working with numbers.
- Being able to identify business problems.
- Being able to work on the given data set.
- Having confidence in learning any new programming language.
- Being able to analyse data from different perspectives.
- Being able to build an ML model for visualising and predicting the outcomes.
- Conveying the insides of particular data to the stakeholders.
- Having good modelling skills, communication skills, analytical skills, and technical programming skills.
If you have the ability to analyse data and extract meaningful information from it, you will fit in a data science team. However, learning basic programming skills, including R, Python and SQL queries, can always be advantageous in this career.
Mixed opinion from experts
Though experts argue over the need for coding for data scientists, stating it is mandatory to work in the field, many non-programmers with a no-coding background still have glorious careers in data science and programming, and coding is more a skill than a criterion.
Experts who say coding is a must, include Rachael Tatman. In her article for freeCodeCamp, she said that every data scientist should be able to “write code for statistical computing and machine learning.” In his recent blog, Ronald Van Loon, CEO, Principal Analyst Intelligent World, provided a long list of technical skills required for a data scientist. He said there was a need to have knowledge about programming languages that included Python, Perl, C/C++, SQL and Java, plus expertise in SAS, Hadoop, Spark, Hive and Pig.
Some opinions also argue that coding is not necessary for a data scientist. In his blog for Rapid Miner, Tom Wentworth says, “Yes, you can do real data science without writing code.” Today, organisations are also identifying ‘citizen data scientists’, who are generally non-coders but capable of solving complex problems of the organisation.
The non-coders also give reasons why coding is not necessary for them:
The reasons include:
– Common algorithms are already known as they are already coded and optimised.
– Explicit coding is being replaced with drag-and-drop interfaces, like Trifacta and Tableau.
– Data science is becoming more automated with options like Google’s Cloud AutoML or DataRobot, both of which help you find the right algorithm.
– Google has also assured that a data scientist can train high-quality custom machine learning models with minimum effort and machine learning expertise.
– Google Duplex demo has also hinted about the future of AI, where the future data scientist might simply be having a conversation with a machine rather than coding one.
A decade ago, very few people used the job title “data scientist”, and there was no such thing as a “data science” degree. The internet was still nascent, and coding was a must; however, there are many algorithms already worked out and universally available online, so coding does not seem to be a “must” anymore.