THIS! I agree with all the points shared here. The model is not important, training processes remain the same. Its the data engineering and you knowing where to look and what to infer is what the company will pay you for.
Defining target variables and stating often what they are. The business side are often all over the place seen from the perspective of the machine learning engineer.
These videos are very useful! Honestly, I would appreciate it if you went into more detail about networking, since it’s by far the thing I struggle the most on. Or at least, if you can point me towards some other resources about it?
As a data scientist every point you made in this video is spot on. I just have one nitpick, when you talk about documentation, the example you show is not how you should do it. Your code should be clean enough by itself to understand exactly what is happening, and you should only add comments that explain why you’re doing something a certain way. To me, documenting is a combination of clean code, adding “why” comments when there are multiple solutions, and writing clear and concise docstrings.
It is so hard to feel connected to the coding community because it seems as if 90% of them are only in it for jobs and don’t really have a passion for it.
I'm stuck, don't know know what to study, where to start from. Completely new to this domain. Any suggestion or direction will be highly appreciated! Thanks and regards.
@@LoveYou-yu8wjyou can always just follow your interests and not think about jumping through hoops to get a job. What made you want to get in to a role in data in the first place? What things interest you outside of data?
But the term “Variance” in the Bias-Variance tradeoff DOES represent variance! It is the variance of the model’s predictions, Var(f~(x)). As the training data X is a random variable, so is the learned model and the predictions it produces f~(X). The rough intepretation of the variance term would be that the model changes a lot based on the training data
14:05 as a developer my PR review would be to remove those comments. Function descriptions are fine. Comments on very complex logic is also good. That is just repeating the code.
2:34 I think learning the assumptions of the models gives already an understanding of when to use or not based if the data is violating the assumptions as well you can test some hypothesis testing and check the distributions to understand better if that's the model you should use but as well the goal of the project will kinda influence but I would say it's more the data that have bigger impact in the moment of choosing a model. I think it's important to know the math but I think to build a model from scratch if we talking about ML problems not DEEP LEARNING i think to make a model from scratch is really forna specific case where pyspark or other frameworks are not ideal what generally are. Im a fan of sickit learn but we all know that deploying models in sickit learn is not the best since the are use for small and medium datasets or for research. But I would say thatsickit learn is the best framework to master a lot of concepts and tools for ml.
i am looking for a job in ml ai , i am a good python programmer , what certificates should i get ? , so far i have roadmap to get harward's intro to ml , google's ml and aws's devops certificate
@wintutorials2282 what subject is not a science? Even history has no point in decorating the events if you don't understand why the events unfolded as they did. And you gotta study the basic major events such as WW2 to understand Cold War
@@MatheusLB2009 hmm yeah I meant more STEM subjects, bc STEM subjects are built on axioms, which make up the fundamentals. So understanding them is required to actually understand without abstraction.
But how can I establish a good math intuition because I am pretty bad at understanding math because my basics are not clear and if I try to learn basics of maths from scratch it will take too much time until I reach deep learning.
Hey would it be possible to contact you for some Questions or even an Interview? Me and a fellow student are working on a study for a guide for AI implementation in the Maintenance field. We would really appreciate your input on some topics regarding the AI field. Best wishes from Germany!
in your opinion, is that worth to learn a data science field? but after watch this incredible video I just realized how complex data science is, and i think I would learn this interesting field. (sorry for my English)
In my research and what I have heard a little bit of understanding of DSA is required to be a "Good Programmer". We as aspiring data scientist shouldnt be unaware of concepts from branching domains.
There is no correct answer to this! While in an ideal scenario it is not required, most companies still screen via DSA/ Leetcode. And knowing atleast the algorithms and what exists can lead to better intuition while programming. Note that you dont only need to build these models, you'll also need to implement them with your existing product or service. I have used data structures numerous times as an AI engineer.
for those watching and reading - I never finished HS - I didnt attend Uni. I have been a paid programmer for almost 25 years now. I never had impostor syndrome, because I never had all the accolades you guys get before starting. Dont believe it - its WEAK thinking. Push those obtrusive thoughts out of your head and just do it. Its the difference between "doers" and "talkers" - "winners" and "losers" I work in AI/ML now - still no degree. I learn the maths I need at each stage.
I strongly disagree on your screenshot with comically obvious and redundant comments as an example of good documentation. Maybe it was necessary 50 years ago, when languages and compilers were not that great, but today it's just like using a horse for transportation. Writing the same thing as the code, only in improper language like English, is bad. It's acceptable as training wheels, but after that it just creates extra confusion and debt. Having a comment is like crying "wolf!", if every line has them, nobody will pay attention to them. It also enables bad naming and structuring practices, since it gives a false sense of clarity. I'd say that if you have an urge to comment something, probably you should rename your variable or extract a function, not just write a comment saying that your variable myCatIsGreat is in fact just an index.
13:59 This is horrible! Documenting all lines of code means you create unreadible code :) Code should be self-explenatory and comments should help to understand main ideas, algorithm features and use cases and sometimes why it is written in this manner (optimization, special case, etc) :D
hey, just started with ML recently, do you have a discord server or insta? would love to connect and ask you a few questions, also one more question, can you please share some examples of real world problems/datasets that can be used to make some decent ML projects? Thanks you!
Thank you sir, it's rare as gold to find such an honest And useful tutorial in a "learn in just X hour" era
This deserves 100x more views - it pretty much lines up with my thoughts on the field, worked out through trial and error.
This channel is a gem! Greatly appreciated!
THIS! I agree with all the points shared here. The model is not important, training processes remain the same. Its the data engineering and you knowing where to look and what to infer is what the company will pay you for.
Anywhere to start for beginners ?
For budding data scientist
Defining target variables and stating often what they are. The business side are often all over the place seen from the perspective of the machine learning engineer.
@@nkristianschmidt thanks
So glad you mentioned the math or getting basic intuition about the math involved.
These videos are very useful! Honestly, I would appreciate it if you went into more detail about networking, since it’s by far the thing I struggle the most on. Or at least, if you can point me towards some other resources about it?
As a data scientist every point you made in this video is spot on. I just have one nitpick, when you talk about documentation, the example you show is not how you should do it. Your code should be clean enough by itself to understand exactly what is happening, and you should only add comments that explain why you’re doing something a certain way. To me, documenting is a combination of clean code, adding “why” comments when there are multiple solutions, and writing clear and concise docstrings.
one of the best video ever.
really I'm greatfull to you because I know this earlier 😊🙏
Wanted to tell ya. This is a great video! Great information. Thank you 🎉
Im starting an ML project in university as someone with not too much experience. This is so helpful
Great and informative video!
It is so hard to feel connected to the coding community because it seems as if 90% of them are only in it for jobs and don’t really have a passion for it.
+
I found this extremely useful as a framework to better understand the domain. Thank you!
Thanks for this! I'm currently self studying the skills needed for DS. Hoping to shift by next year.
Same here
@FlotsamDM
@AltraPowerYT
I am also self studying DS
I would be cool if we could connect via discord or something
I'm stuck, don't know know what to study, where to start from. Completely new to this domain.
Any suggestion or direction will be highly appreciated!
Thanks and regards.
@@LoveYou-yu8wjyou can always just follow your interests and not think about jumping through hoops to get a job. What made you want to get in to a role in data in the first place? What things interest you outside of data?
This is solid life advice, though a bit daunting
wow this is insanely compact and useful video about ML that people underestimated !!!
Great!! and Informative video
But the term “Variance” in the Bias-Variance tradeoff DOES represent variance! It is the variance of the model’s predictions, Var(f~(x)). As the training data X is a random variable, so is the learned model and the predictions it produces f~(X). The rough intepretation of the variance term would be that the model changes a lot based on the training data
All he said is pure facts. Specially the parts of fundamentals!
is it machine learning or motivational speech?
Excellent points every single one of them
Excellent video
14:05 as a developer my PR review would be to remove those comments. Function descriptions are fine. Comments on very complex logic is also good. That is just repeating the code.
Great job 👏
Thank you for this video.
gained a lot..thanks
@3:36 where is this image from? :-)
2:34 I think learning the assumptions of the models gives already an understanding of when to use or not based if the data is violating the assumptions as well you can test some hypothesis testing and check the distributions to understand better if that's the model you should use but as well the goal of the project will kinda influence but I would say it's more the data that have bigger impact in the moment of choosing a model.
I think it's important to know the math but I think to build a model from scratch if we talking about ML problems not DEEP LEARNING i think to make a model from scratch is really forna specific case where pyspark or other frameworks are not ideal what generally are. Im a fan of sickit learn but we all know that deploying models in sickit learn is not the best since the are use for small and medium datasets or for research. But I would say thatsickit learn is the best framework to master a lot of concepts and tools for ml.
Very good welldone!
Hey, why you put such obscene pictures in the starting?
Game changer
I appreciate the correct spelling of impostor.
I have seen the word 'impostor' being spelled as 'imposter' so often that I started thinking that the ladder was the right spelling
@ haha yeah same, I only know because I googled it after seeing the word in Among Us, thinking they had spelled it wrong
british vs american english who tf cares go learn data science.
i am looking for a job in ml ai , i am a good python programmer , what certificates should i get ? , so far i have roadmap to get harward's intro to ml , google's ml and aws's devops certificate
Understanding math will make you a better human being in general.
bro starting off with image of person wrist deep into the backside of animal lmao
You should always focus early on the fundamentals, regardless of which field are you studying, period
Nooo not at all. Only sciences. Other subjects not so much
@wintutorials2282 what subject is not a science? Even history has no point in decorating the events if you don't understand why the events unfolded as they did. And you gotta study the basic major events such as WW2 to understand Cold War
@@MatheusLB2009 hmm yeah I meant more STEM subjects, bc STEM subjects are built on axioms, which make up the fundamentals. So understanding them is required to actually understand without abstraction.
I don't find the video of the math skills that make machine learning easy(and how you can learn it )
But how can I establish a good math intuition because I am pretty bad at understanding math because my basics are not clear and if I try to learn basics of maths from scratch it will take too much time until I reach deep learning.
Check out my previous video on that subject: ruclips.net/video/wOTFGRSUQ6Q/видео.html&ab_channel=InfiniteCodes
Hey would it be possible to contact you for some Questions or even an Interview? Me and a fellow student are working on a study for a guide for AI implementation in the Maintenance field. We would really appreciate your input on some topics regarding the AI field.
Best wishes from Germany!
I use SPSS for machine learning.
Claude and chatgpt make data processing and training model vey simple, even i don't know nothing about it,note: i am saying for college level projects
in your opinion, is that worth to learn a data science field? but after watch this incredible video I just realized how complex data science is, and i think I would learn this interesting field. (sorry for my English)
This guy is speaking facts
what is meaning of MLops should a fresher learn MLOPS
I m gonna show this to every DS/ML new grad I get
ive been watching your videos they are really helpful tho can you suggest me is dsa important for ds?
In my research and what I have heard a little bit of understanding of DSA is required to be a "Good Programmer".
We as aspiring data scientist shouldnt be unaware of concepts from branching domains.
There is no correct answer to this! While in an ideal scenario it is not required, most companies still screen via DSA/ Leetcode. And knowing atleast the algorithms and what exists can lead to better intuition while programming. Note that you dont only need to build these models, you'll also need to implement them with your existing product or service. I have used data structures numerous times as an AI engineer.
for those watching and reading - I never finished HS - I didnt attend Uni. I have been a paid programmer for almost 25 years now. I never had impostor syndrome, because I never had all the accolades you guys get before starting. Dont believe it - its WEAK thinking. Push those obtrusive thoughts out of your head and just do it. Its the difference between "doers" and "talkers" - "winners" and "losers"
I work in AI/ML now - still no degree. I learn the maths I need at each stage.
Can you recommend what to learn first for beginners
This is so good so rich
I strongly disagree on your screenshot with comically obvious and redundant comments as an example of good documentation. Maybe it was necessary 50 years ago, when languages and compilers were not that great, but today it's just like using a horse for transportation. Writing the same thing as the code, only in improper language like English, is bad. It's acceptable as training wheels, but after that it just creates extra confusion and debt. Having a comment is like crying "wolf!", if every line has them, nobody will pay attention to them. It also enables bad naming and structuring practices, since it gives a false sense of clarity. I'd say that if you have an urge to comment something, probably you should rename your variable or extract a function, not just write a comment saying that your variable myCatIsGreat is in fact just an index.
this photo exactly my brother did (was vet) :D
Source camera wind.
12:41
Good one tho
14:18
13:59 This is horrible! Documenting all lines of code means you create unreadible code :) Code should be self-explenatory and comments should help to understand main ideas, algorithm features and use cases and sometimes why it is written in this manner (optimization, special case, etc) :D
besides the valuable info of this video, its the first time i am actually laughing on a sciense video
99% accuracy in what and how? 😂
Mb 99% classes displace
00:01 wtf
I don't understand if this is aimed at juniors or people who are still learning the fundamentals. I feel like I learned nothing new tbh
Wtf is this 0:01
aitutorialmaker AI fixes this. Machine Learning Lessons by Infinite Codes
hey, just started with ML recently, do you have a discord server or insta? would love to connect and ask you a few questions, also one more question, can you please share some examples of real world problems/datasets that can be used to make some decent ML projects? Thanks you!
17:04