DataCamp Competitions—10 tips from the Workspace Competition’s winners
Taking part in a data science competition can be challenging. Not only is there completing the task itself to think about, but there’s also the extra question of how you can impress the judges and crowds enough for your publication to win. Additionally, since your publication will appear on your user profile, you’ll want to ensure it will impress anybody who may see it afterward—including potential employers.
That’s why we reached out to some of the recent Workspace competition’s winners to get their advice on how you can repeat their achievements—from ways to approach the exercises and make them easier for yourself, to sprucing up your entry at the end.
Read the judging criteria. For example, creativity (i.e., "How 'out of the box' the analysis conducted is") was one of the highest weighted criterion in the Workspace competition, and the "analysis" criterion also mentioned that "Visualisation tools and techniques utilized" carry some importance; hence, I made a conscious effort to use creative visualizations, whenever applicable.
Templates are your friend. Find an existing DataCamp course, project, or workspace template and use it as a basis to build your solution upon. More or less, there are standard processes in EDA and you just have to adapt them to the specific problem.
Talk to your readers, since this is an activity that will be published on DataCamp, that you'll feature in your portfolio, and that you'll share with colleagues and business professionals.
Don't hesitate to try stuff out. Your first idea will likely not be your final idea, so use this opportunity to test something you've been wanting to test, implement something you've been wanting to learn, etc.
Think outside the box. It’s easier said than done, but it can be a good idea to take the competition’s instructions a bit further. There are sometimes clues within them: in my case, the description asked to find "what kind of people get which kind of loan”. This spelled "clustering" to me: to divide the loaners in groups and find the group characteristics. With hundreds of submissions, you’ll want to stand out.
Take time on your visualizations. Allocate time to perfect the presentation elements of your publication (graphs, outline, final render on the DataCamp publication system, etc.).
Have fun with it: Find an aspect of the problem that you'll enjoy doing the analysis on. For example, I love applying methods on (unstructured) text-based data, hence I often have a tendency to choose independent variables that are text-based, and then work with those in my modeling.
Think about what your research objectives might be. Then, think about what the dependent variables and independent variables may likely be. There's often a choice to be made here, contingent on the research objective you decided to pursue. The best kinds of objectives are when your own curiosity is fully engaged, which will help drive your analysis and make things interesting.
Remember the overall goal of data science. Every step in the analysis, each decision you take, should be with the business problem in mind.
Don't submit at the last minute—in case there are any issues with publishing. After all that hard work, you wouldn't want to miss the deadline due to a technicality.
Many congratulations to all four of them on their achievements, and if you’re feeling inspired check out our current and upcoming competitions!