The data science take-home exam could prove daunting. Some are inherently good at doing the take-home exam. However, others excel at an in-person test, an interview or even a structured programming test than a take-home. Regardless of your personal strength, you have to make sure that you put your best foot forward at the next take-home exam. With a combination of careful preparation, attention to detail, and a willingness to learn from any mistakes, you can increase your chances of success in this competitive field.
1. First things first. Prime yourself.
One important way to approach the problem is to prime yourself into the right mentality. This is especially important if the position is a highly coveted one I remind myself that I will not provide any reasons for the evaluator to reject my application. It is your choice to pick a simple phrase to tell yourself before sitting in front of your keyboard. This will help you perform to the best of your abilities at your next take-home exam.
2. Shift your point of view to the evaluator
The age-old adage, “To catch a thief, you must think like a thief,” holds true in many aspects of life. Whether it’s tracking down a criminal or striving for a coveted job opportunity, understanding the mindset of one’s opponent is crucial. In the context of a job application, it’s essential to remember that the person reviewing your application is envisioning you in the role you’re applying for. Show that you are capable of critical thinking and your decisions are supported by evidence. Submitting subpar work will only lead them to believe that your performance on the job will be equally lackluster.
3. Know your basics.
It is likely that you have had some initial conversations with the prospective company. Make sure to ask for the expectations for the take-home exam at your conversations. You will get a variety of problems from a simple visualization, data exploration to proposing a pricing strategy for a product. Although it is difficult to forecast what you should follow in every project, here is a checklist for you to cross off if the test is a machine learning one.
1. Thorough Data exploration and visualization
2. Data cleaning
3. Checking for class imbalances
4. Normalization
5. Feature engineering
6. Feature selection
7. Model selection
8. Appropriate metrics based on the model results
4. Make it readable.
Picture yourself narrating a story to a child who wants to be put to bed. That is how you should narrate the project from importing the data to generating the insight. It is the art of storytelling through clever use of visualizations and carefully generated statistics. Whether the evaluator reminds you specifically or not, the comments have to be in place explaining every step. Given that there are ample resources online to get coding, it is important to let the evaluator know that you understand every line of code and how it contribute to your overall objective.
5. Never send it away in haste
I worked on a project that consumed a lot of my free time. This was shaping up very nicely as it was a particularly intriguing topic for me. However, I decided to send it while I was having a conversation with a friend only to discover that the sent report was riddled with many formatting errors. It was not only a lesson for data science projects, but certainly a one for life.
6. Resilience
In these uncertain times, it is crucial to understand that the job market for data scientists is highly competitive and challenging. Many companies have been forced to make difficult decisions, resulting in layoffs and limited job opportunities. If you encounter rejection, use it as an opportunity to learn and improve your future applications. Though it may be disheartening, it is important to remember that feedback on the project may not always be provided. It is essential to approach these setbacks with a level-headed and resilient attitude.
