Gearing Up For Your First Data Science Interview
“Data Scientist: The Sexiest Job of the 21st Century” so read the headline of an article in Harvard Business Review (HBR). This article alone explains the newfound craze to make a career in data science. The factor that it pays quite well as well is like adding a cherry on the cake.
But how do I get into one? I get asked this question a lot by my friends and other folks. I can understand their confusion as this field is still getting its shape. The top academic institutions themselves have started to roll out degrees in data science quite recently. This has made the field filled with candidates without degrees and with no theoretical knowledge.
There are of course 2 ways to look at it. First, this is too confusing for me so I am out. Second, the field has a few degree-holders, constraining companies to entertain the rest too. So, if you prepare right, you too can land a job!
So, how do you prepare for your first interview? Let me introduce you to the “Arsenal of a Data Scientist”. Yes, I myself coined this term!
1. Learn Python as the ‘programming language’
Yes, there are other emerging languages but Python still rules the programming billboards. You should focus on learning only this language and get good at it, and by good, I don’t mean getting good like coders working at Google. Learn and practise the basics such as conditional statements, loops, and classes. Even classes are rarely used at work. When it comes to data structures, the inbuilt ones in Python such as dict, list, and tuples should suffice. There is no need to learn the heavyweight stacks, trees, and graphs etc.
2. Practise enough programming problems and logical puzzles
There are a lot of portals online dedicated to this. Pick any credible one out of them. In my experience, geeksforgeeks.com works the best. I have seen their questions in the interviews themselves (in India) many times. Practice till you get confident of strangulating a given string in any direction you want!
Moving on. puzzles, they too are very common in interviews. Practise about 30–40 puzzles from geeksforgeeks.com and you are good to go. Remember them by heart as well since they too can be asked in the interview too.
3. Learn SQL for querying data out of databases
Learning the whole SQL shouldn’t take more than a day or two. You can use the good old w3schools.com site. Focus on the joins — inner join, left join, outer join, and cross joins etc. Practice it on a couple of problems on w3schools.com or a different site and you are good to go.
4. Learn topics required to run an ML project till its end
In case you come from a non-statistics background, you will need to invest a lot of time in learning the theory part. Since there is a lot to cover it becomes overwhelming for some. I suggest that you first start with very few topics that allow you to start getting your hands dirty on ML projects. These topics also cover over 80% of a typical data science interview for an entry-level role. I, after going through over 30 interviews, have compiled a list of them. I have published it on my site ml-concepts.com for everyone to use. Don’t leave any SINGLE topic on this site. You will find that everything else revolves around them more or less.
5. Practise Practise Practise
Finally, we come to the implementation part. You should pick 2–3 good ML projects on Kaggle.com. I too will add some important projects on the site targeting interviews. Implement every project all by yourself line by line. Learn each line and its role.
That’s all! These points should get you started for your first data science interview. Best of luck!!!
You can reach out to me at sourabharsh@gmail.com