Original Source Here
Bias and Barriers
It is tempting to jump with both feet into a Kaggle Notebook that offers a solution, a ready-to-use baseline model, and a canned submission. However, being able to have an answer in no time is as enticing as it is dangerous.
Following the easy way is a trap. It forces us to make the same assumptions as of the Notebook author and embrace the same biases. From this point on, our creativity will suffer, and we will quickly find ourselves making circles around the same ideas and concepts.
In the end, the only parameters of the problem that we will be able to revisit and change are minor things, such as the number of layers in a Neural Network or the value of the Learning Rate. This methodology can only get us so far; in the best-case scenario, our solution will be a few spots above the leaderboard.
In my opinion, peeking into off-the-shelves solutions does not make us better in any way. It does not let us build our understanding, form our assumptions, compile our code and even fail miserably. This is what it is all about; a satisfying and fulfilling learning process.
Fertilizers and Catalysts
So, should we start working on our own, in a silo, designing and running our experiments before even taking a look at other’s work? Not at all; it’s like arguing that composers should not listen to music because they might get influenced, or authors should not read books in fear of writer’s block.
If I have seen further it is by standing on the shoulders of Giants. — Issac Newton
There are a lot of great Data Analysts on Kaggle we can learn from. The learning process is not done in a vacuum. We need to spot their work. But, what makes a Data Analyst great? I argue that there is only one gauge for judgment: asking the right questions.
The Notebooks we search for here are provoking, stimulating, and heretic. They trigger our imagination and creativity by asking helpful questions.
An excellent Data Analyst provides us with a green field, a horse to ride, an idea, and the tools to conquer it. Even if we disagree with their findings or suggestions, we have already taken a step forward.
A great Data Analyst produces inspiration for us, bringing the data or features that seem promising and relevant to our attention. They transform the datasets in such ways that highlight potential. And in the end, summarize the core points fast, and compress them into easily memorable chunks.
Great data scientists are heretics. They stimulate our imagination and provoke our creativity. They shape the raw material in a way that highlights the parts that matter. In a nutshell, they ask the right questions for science to solve.
Next time, when competing in Kaggle, search for those Kernels. You’ll grow more even if you don’t submit at all.
About the Author
My name is Dimitris Poulopoulos, and I’m a machine learning engineer working for Arrikto. I have designed and implemented AI and software solutions for major clients such as the European Commission, Eurostat, IMF, the European Central Bank, OECD, and IKEA.
Opinions expressed are solely my own and do not express the views or opinions of my employer.
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot