Data science is a lot like cooking. Although raw ingredients may be fascinating at first, the fun doesn't start until you're actually able to start slicing, dicing, and eventually serving up something delicious to devour. Most of the time, you'll end up with a dish, but in the data science world, we call it data insights. In this blog post, I want to share 5 analogies of data science to cooking which helped me understand the field better.
1. Without good ingredients, you can't cook a good dish
Just like ingredients, data are the raw materials. And as the saying goes, "garbage in, garbage out". Good data is key to a successful data science project. The output of your machine learning model is just as good as what you put inside it. Hence, it is important to make sure that your data contains enough relevant features and not too many irrelevant ones. Without relevant and quality data, you can't really do any useful data science.
“Data are becoming the new raw material of business.” — Craig Mundie
2. Most time and effort are spent on cleaning and preparing the ingredients
The fact that most time and effort in cooking are spent cleaning and preparing the ingredients will resonate with anyone who's had a helping hand in the kitchen. This is also true for data science. But instead of slicing, dicing, and marinating; we have feature engineering, data cleaning, and normalization. Cleaning and preparing the data is required before delivering insightful visualizations and analytics that can eventually drive data-informed business decisions.
“Data science is 80% preparing data, 20% complaining about preparing data.”
3. Different tools and techniques are needed for different recipes
A cozy meal for two requires different tools compared to catering for 2,000. Similarly, processing 1,000 rows of data may run on a laptop, but processing a billion rows may require specialized distributed systems and servers. Choosing the right techniques is also important for both of these tasks. No one likes an undercooked or overcooked dish, just like underfitting and overfitting in data science.
“There is no super algorithm that will work perfectly for all datasets.” — No Free Lunch Theorem
4. Cooking is both a science and an art
Just like cooking, you need certain tools and techniques, but you also need creativity and intuition. Data science doesn't exist in a vacuum, it must relate to other areas for it to have the greatest impact. Packaging the numbers creatively in a way that can be interpreted by others is crucial to getting them to see the whole picture and therefore finding the best solution. When you approach data science in a creative way, the results are often astonishing.
“Talented data scientists leverage data that everybody sees; visionary data scientists leverage data that nobody sees.” — Vincent Granville
5. You can't become a great cook overnight
You can cook something by watching a video or reading the recipe from a blog. But, it doesn't really make you a great cook. Similarly, you can do some data analysis or modeling by copying code, but that won't make you a great data scientist. Likewise, completing a course or earning a certificate won't make you a great chef, and it won't make you a great data scientist either. It takes years of dedication, effort, and practice. Data science is a journey, not a destination.
“Learning data science is like going to the gym, you only benefit if you do it consistently.” — Moez Ali, Creator of PyCaret