What are the most in demand Data Science skills?
Knowing what to learn or focus on is a key part of improving as a data scientist or getting into the field in the first place.
Data Science is a fast growing field with many in demand skills that one could learn, however it can be difficult to decide on how to prioritise. Should I focus on deep learning frameworks or statistics, for example. That’s what I hope to help with in this blog post.
Data science is a large enough field that the skills required to be a ‘data scientist’ can be very different depending on where you want to work. So I’m going to break these down into two areas: what is needed to work in big tech companies (read: Netflix, Spotify, Facebook, Google etc), and what is needed to work in a small or medium sized company (that has a data science team…).
What data science skills are big tech companies looking for?
Large ‘digital first’ companies have understood the importance of data science a lot longer than most, and are thus ahead of the curve. So, if you want to see what the future of data science will be and the skills you will need, look at their job postings and what they are looking for in candidates. These companies will often have separate teams dedicated to data science, data engineering, and machine learning. This results in roles that are more specific and focussed on a particular area. It is therefore quite common to find two roles relating to data science within these companies:
Data scientist
- Finding insight
- Analysis
- Experimentation
Machine learning engineer
- Developing machine learning models
- Operation and optimisation of models
This distinction clearly indicates that machine learning and data science are undertaken by two different people. A data scientist within these large companies will be tasked, instead, with finding actionable insight from large amounts of data, understanding customer behaviour, and running experiments to help inform product decisions. The creation of machine learning models, conversely, is handled by a different team… this distinction is becoming more common in the job market, so make sure you're ahead of the curve! Therefore, the skills that are actually required from the ‘big tech data scientist’ are centred much more around statistics and data analysis than they are machine learning.
So, if this is the route that you would like to go down then I suggest that you focus on the following:
Statistics
- Experimentation
- Hypothesis testing
- Confidence intervals
- A/B testing
- Distributions
Data analysis
- Fluency in SQL and Python
- Knowing how to manipulate data
- Transforming data
- Dealing with outliers
- Time series analysis
- Finding and presenting actionable insight and outcomes in an effective way
A lot of these skills are often not the sexy skills that data science courses and news articles talk about, that is much more focused on deep learning and ‘AI’. However, these are the tasks that a data scientists needs to be able to do blindfolded with both hands tied behind their back to be considered for a data scientist position at a ‘FANG’ type company.
Being a data scientist at a small or medium sized company
Most roles within a smaller company require you to wear multiple hats, the same goes for a data scientist. In contrast to the big tech position, a data scientist in a smaller company will be required to cover both the experimentation, analysis and the machine learning engineering. In extremely small teams you may also be required to undertake data engineering tasks.
This means that the skillset of this data scientist would need to cover all the previously listed areas and additional machine learning responsibilities:
Statistics (as listed above)
Data analysis (as listed above)
Machine learning
- Machine learning model development
- Model deployment
- Model maintenance
- Machine learning operations
The wide range of skills required in a smaller company gifts the opportunity of developing an impressive set of skills which could be useful for positions and situations later on in life, however it also means that focus is split and you may not gain the deep understanding and experience of ‘core’ data science concepts.
What you need to know to get any job in data science
Now that we have gone through the skills needed and the differences that exist between different sized companies in data science, it is clear that the answer to the question “what do I need to know” is… “it depends”. However, there are some cornerstone skills that will put you in a great position to get a job anywhere in data science, these are:
- Understanding how to analyse and work with large amounts of data
- Finding actionable insight from datasets
A solid statistical understanding of:
- Experimentation
- Distributions
- Effective communication of insight
Without these skills as the foundation, getting a job in data science will be incredibly difficult. You can create the most complex deep learning model imaginable, but if you don’t know how to run an A/B test then you won’t make it past the screening round of the interview.
Resources
If you would like to strengthen your knowledge of the skills required for data science then I recommend checking out the following links:
