2020年9月10日星期四

HKU Webinar - Winning on Kaggle: Becoming a better Data Scientist

The HKU Webinar named “Winning on Kaggle: Becoming a better Data Scientist” was organized by AI Lab and MEICOM and held on 10th Sep 2020.  Prof. Qingchen Wang (HKU Business School) was the guest speaker.  In the beginning, Mr. Alan Chow (Founder, AI Lab) briefed that AI Lab focused AI, Blockchain and Cybersecurity, etc.


Firstly, Prof. Qingchen Wang introduced four goals of today’s talk including introduction of Kaggle competition, their framework and tips for this competition and demonstrated some real competition cases.


Kaggle competition had started since 2010 and focused on forecasting, image recognition, sentiment prediction, stock prediction, product search, predictive maintenance, etc.  Then Prof. Wang said “Kaggle is a grind, but you learn and gain in real life from it.  Kaggle will definitely make you a great data scientist.”

However, some people said “Why Kaggle will not make you a great data scientist” because of mainly focusing about prediction.  Some claims were missing the link to return on investment (ROI) analysis. And then Prof. Wang advised to be an excellent data scientist that everything else is a matter of experience. 


After that Prof. Wang briefed what would learned from Kaggle such as data science is rigor, understanding your model, 80/20 rule of data science, the power of team work, determination and grit is most important, etc.


Then Kaggle details inform was briefed and benefits were stated.


Finally, Prof. Wang suggested students that data science is their career, think of everything you done now as an investment for next 20-30 years. Kaggle could learn something out of scope in university.


At the end, Prof. Wang encouraged student to get start in five steps.
1.          Understanding the prediction problem.
2.          Understanding the data
3.          Read the discussion forums
4.          Start with a public notebook
5.          Develop your own solution.


Reference:

沒有留言:

LinkWithin

Related Posts with Thumbnails