It’s a Friday evening and I’m glued to my laptop while everyone else is probably partying. I come across a very interesting LinkedIn profile of a Data Scientist who works for LinkedIn. I go through his posts and find one particular discussion interesting. The discussion is if an advanced degree(Masters/PhD) is a necessary qualification for a data scientist.
I notice that a very good friend of mine has commented on the same, favoring MOOCs and Kaggle competition performance over universities. With new courses on Deep learning by Coursera and high competition on Kaggle, the industry standard are clearly shifting towards knowledge from MOOCs and use of more efficient packages such as H2o and ensemble such as xgboost and adaboost.
I tried asking people from academia and industry about what companies look for in data science professionals and the findings were very interesting. While Kaggle competitions give great visibility to participants especially from the most leading data science companies, many of the other companies don’t care a lot about advanced package you use. They care about the fundamental knowledge of statistics and enough domain knowledge to identify how to handle data and what features to engineer upon. The data science market is still young and there are many startups struggling to find quality data scientists. For these companies, it doesn’t matter what packages you use. They care about how you can apply concepts and how reliable you are in delivering a solution. There are also companies that have dilemma about what weightage should be given to statistics and domain knowledge. One quality every companies irrespective of size is looking for, is good communication and soft skills.
Yes, different companies have different criteria for selection of data scientists, however there are some common factors that all data science practitioners need to work on namely, good knowledge of statistics, good communication skills and great domain knowledge. Improvements in these areas should land you your dream job according to current practitioners in the industry. MOOCs are a great way to acquire knowledge, but universities are also doing a great job at imparting knowledge and teaching the fundamentals of statistics. It is left upon each professional how he makes the best of the opportunities given to him/her by his/her source of education.
Just putting across my thoughts and findings. Feel free to leave a comment