Stop with the Kaggle Nonsense

We don't use Kaggle to make hiring decisions.

[jaredgreathouse]

Speaking as a data scientist in academia, getting your data together is CERTAINLY the hardest part of any report or project. In real life, rarely do we ever enjoy the luxury of clean data being handed to you. Sure sometimes you do! And we're thankful for those moments. But this is decidedly NOT the norm in any sense of the term, so the description of Kaggle being for playtime not prime time is pretty damn true.

[mustafayahya]

@datajanitor. Sir i understood what you meant and it is really true. If i learn recommendation systems which use SQL then will it be right to start learning this?

[Rafael-xujh]

if kaggle is not reliable, where should I practice data science while I'm still trying to land my first job as a data scientist?

[philosophia]

Why in the world I'm I seeing your videos just now after so many months of wasted time on bullshit certs and courses.? I'm coming for your training Sir. Never come accross anyone like you in the data sphere Like even the terminologies you use like "data roles", "data analyst" etc, there's just so much intention and specificity behind them.The sense of direction you give me is most appreciated. max respect!!

[ethanblackthorn]

Thanks for the video! Can I ask your opinion on whether Kaggle is a good place to practice making good visualizations and well-written code with comments instead of trying to reach that top 1%?

[Muan]

I agree with you about having a ready clean data on Kaggle and building best score is useless but people can learn machine learning application to data. So if I cannot use Kaggle projects as experience then I need to find my own projects I guess

[navidjavdani]

I'm feeling lost and confused. It seems like data cleansing is the crucial aspect of being a Data Analyst, comprising about 90% of the job. However, you say that modeling data is only around 10% of the role.
To increase my chances of being hired by a company as a data analyst or in machine learning, what specific skills should I focus on? Where should I channel my efforts to enhance my employability?

[mattiasfagerlund]

I used to write my own machine learning algorithm implementations (in c#) to compete in Kaggle - nowadays you're only allowed to use python notebooks for most of the competitions. So I'm stuck using the same tools as everyone else, which takes the majority of the fun out of it for me. Also, 90% of the top of the board are people copying someone elses notebook and simply changing some random seed. I guess they want bragging rights for being high on the list. For my money, Kaggle used to be waaay better, but it was always extremely artificial and people will always game their entries to do good on the test even if it means doing poorly in a real application.

[hxxmiso]

This is real, am in a competition where my model is 0.42.. accurate yet am in top 10, when I fine tune for accuracy my ranking drops in kaggle competion

[adrian]

So what in your opinion is the best way to get intuitive expertise in ML? I mean the state in which vast majority of your coding you don't google or chat but just do it from your head (and you do it right). Probably becoming ML engineer is the only right way to get there but do you have in mind any other way? Would it be self-created, tasted, deployed web app based on ML with focus on quality, some quick models in colab with focus on quantity, maybe some alternatives for Kaggle or instead doing projects it's better to learn theory, read documentations and don't waste time on coding something that I don't fully understand?

[ko-Daegu]

As a web developer can I jump straight to ML Eng ??
In Bachelor in CS we did a lot of Probability theory and statistics so I feel I got the foundational math needed
What do u think ??
Or Data analyst is a must for ML Eng

Do I buy your ML Eng and go with that or with Data analyst ?

[pablomorniroli]

Hi Mike!

If we can't count on Kaggle, where can we find real life databases?

How useful to work with them and make projects for our portfolio.

Thanks for the information you share, I find it unique. You unmask a lot of people in this business.

[utkugulgec]

Guess it is only good for familiarity with the concepts

[RafaelGarcia-kxyt]

Thanks, I thought i was crazy, because Kaggle is such a "famous" platform, but not having to do ETL and also seeing models with 100% accuracy win money... wtf? Anything with more than 95% accuracy already makes me sad, because certainly there IS a problem on the model or ETL.

[ahmedibrahim]

On my resume, I add Kaggle as a hobby not an actual source of experience. I see Kaggle as a ML sandbox kinda what Qwiklab is for Linux or Cloud learning.

[rajshahmat]

What is you take on Database Administrator role for entry level?

[mybiru]

Any resources (if not kaggle) we should consider investing our time??

[erickheredia]

Stop this man now. He's making too much sense, hehe. You have to be careful. DS bootcamps are coming for you.

[georgel]

So where can i find great data scientists to outsource?

[kennet]

Damn back in the days I saw people with top Kaggle as someone to look up to. As I grow and gained experience, software engineering is about manage entropy, rather than pure programming.