Why Kaggle is So OP

Personally, Kaggle is sufficient enough for someone for data science starter, or even to delve deeper in data science field. This article was written after having a conversation with a Kaggle Master, listening to some Kaggle Grandmaster Podcast, and try by myself on Kaggle. What I can conclude are;

1. Complete all the Kaggle mini-course. This mini-course is provided on Learn tab on Kaggle. Most of ml tutorial taught about the text book of DS, but this mini-course taught the technique which is what makes it priceless. Technique is what I feel I had lost comparing with all the expertise, especially on the feature creation and validation methodology where it is really useful for most of the cases.

2. Read the Kaggle forum. This is the gold mine treasure where a pool of Kaggle master and grandmaster having fantastic discussion together. Hate to say that we are not getting some of the knowledge in article such Medium, because they just talked casually here. Other than question answering, the Kaggle discussion also shared their solution which is another precious diamond, with the title, 1st place solution.

3. Join competition. By joining competition, we will understand how Kaggle works. There is two type of competition which is code competition and submission competition. When I first knew about code competition, where we need to turn off the internet, it feels really weird. But that how the competition worked.

4. Listen to the Kaggle master and grandmaster talk/podcast. My friend who are Kaggle master always did this to learn something new and implementing in his code. Last weekend, I listened to Giba, the NVIDIA Kaggle Grandmaster. As a novice, he taught a lot of very useful information, tips and technique that can be applied within the data science project or competition.

There might be more. I will update the article soon or later. (Last update: 15-9-2024)