do i need to know the basics of computer science before getting in to programming languages like SAS, R, PYTHON etc?

chungus

chungus

Market Restricted
Disciple
so i wanna learn data analysis with languages like SAS but i do not have any kind of knowledge regarding computer science and stuff. I tried watching videos by freecodecamp but most of the stuff goes above my head. what do you recommend i should do? i also bought a adruino so that i can practically see how a programming language works but im not getting time to tinker with that. I just want to know basic fundamental like what people teach kids in schools
 
Arduino isn't a bad place to start. The IDE uses C++. Get a few LEDs turning on and off. Write a string to an LCD/OLED display. Make a pong or snake game. Then make a line-following / obstacle avoidance robot etc. It doesn't matter if nobody has taught you the basics. Each time you hear a term you've never heard of, do your best to research it and understand it before moving on. Use ChatGPT/Claude and ask it to explain it like you're 5.

CS is easy, you don't need a university course on it. You learn it on the fly. Just don't bounce from one thing to another, from one MOOC to another and from attempting to learn one language to another. Whatever you choose at first, gain some proficency in it before moving on and attempt your own creative projects with it. That way learning the next thing will be much easier. That is the #1 mistake most people make. They just aimlessly complete basic tutorials until they give up.
 
Not familiar with SAS. I did CS long ago, had a coding job for some years. Today i work with python + pandas for some basic testing for trading professionally.
I can write some simple in concept but pretty extensive projects for my work.

1) I do not remember much from my CS degree. So from that maybe you have some answer, as i code today as well.
2) Coding has its own skill set in being able to write abstract visualize test debug etc. That will come with time and practice, maybe few years. Progress might make sudden jumps ( It did once for me ).
Its not rocket science, best way to learn after getting familiar with syntax is just to keep working on projects. Don't get lost on details, google/stack exchange/similar is your friend as a reference.
3) Do learn to use data structures.
4) python is very easy to pick up and use.
5) You will certainly need to learn parts directly related to what you are trying to do and if that is part of CS then learn that and dependencies. SAS i do not know.

good luck.
 
Last edited:
If you want anything approaching accurate knowledge, do NOT ask chat bots to explain or tell you stuff.

I dunno how you think that's appropriate, vector, to advise a beginner to do that, given how much of a complete shitshow chatgpt's knowledge is at the moment.
 
calvin1719 said:
If you want anything approaching accurate knowledge, do NOT ask chat bots to explain or tell you stuff.

I dunno how you think that's appropriate, vector, to advise a beginner to do that, given how much of a complete shitshow chatgpt's knowledge is at the moment.
Click to expand...
It's very appropriate, especially for a beginner.

LLMs only start to hallucinate when you give them complex problems. There are exceptions to that rule but with the kind of roadblocks that a beginner like @chungus will be running to, a decent LLM will speed up his learning process by a lot. LLMs are excellent at explaining what an array is to you, or walking you through fizzbuzz etc, or helping you set up a dev environment. They're very, very unlikely to fail you there.

I also think of learning as an iterative process. One where you start off with a blurry, pixelated image and gradually increase the resolution i.e the scope and accuracy of your knowledge over time. So even if the LLM occasionally misleads him, which I would argue happens less than 1% of the time for a situation like this, as long as it helps him along much faster -- it is totally worth it.

Also, I use LLMs constantly to save hours of time. I use them to construct linux commands, write quick python scripts for me etc. So I'm advocating something that I also do.
 
Last edited:
I feel like this should be its own thread, are llms good for learning anything at all? Discuss.

Agree to disagree I guess. I don't want to derail this thread with that discussion.
 
variablevector said:
It's very appropriate, especially for a beginner.

LLMs only start to hallucinate when you give them complex problems. There are exceptions to that rule but with the kind of roadblocks that a beginner like @chungus will be running to, a decent LLM will speed up his learning process by a lot. LLMs are excellent at explaining what an array is to you, or walking you through fizzbuzz etc, or helping you set up a dev environment. They're very, very unlikely to fail you there.

I also think of learning as an iterative process. One where you start off with a blurry, pixelated image and gradually increase the resolution i.e the scope and accuracy of your knowledge over time. So even if the LLM occasionally misleads him, which I would argue happens less than 1% of the time for a situation like this, as long as it helps him along much faster -- it is totally worth it.

Also, I use LLMs constantly to save hours of time. I use them to construct linux commands, write quick python scripts for me etc. So I'm advocating something that I also do.
Click to expand...
i have installed llama3 but havent tested it yet but ive tried chatgpt & it cant even create basic html file with css
 
chungus said:
i have installed llama3 but havent tested it yet but ive tried chatgpt & it cant even create basic html file with css
Click to expand...
Odd. I've used it to do exactly that before -- mock up a website with html/css/js. And that was with 3.5, not with the new 4o. You may need to be more specific in your prompts.
 
Here is a prompt I used in GPT 3.5 make a a html file and link a css and JavaScript file that's in the same directory and it gives me working code.
I don't see how GPT can **** up with basic code unless you ask it to build you a $100b startup.
Try to be very specific, and this is where knowing a bit of coding comes it.
 
Last edited:
like other's said, LLMs are good for basic stuff but if/when you venture into an atypical use case it'll show its limits, that said it's fine to use it as a supplement as long as you follow up with traditional articles, we are heavily leveraging LLMs in our org for semantic parsing and its something else in that usecase. for the OP, you can definitely use chatgpt to break down concept for you but make sure you follow up with some articles/documentation just to make sure its not hallucinating, thats how I helped my SIL learn Python/SQL without me handholding her throughout the process
 
chungus said:
so i wanna learn data analysis with languages like SAS but i do not have any kind of knowledge regarding computer science and stuff. I tried watching videos by freecodecamp but most of the stuff goes above my head. what do you recommend i should do?
Click to expand...
Get a tutor for SAS. That's what you want to learn.
chungus said:
i also bought a adruino so that i can practically see how a programming language works but im not getting time to tinker with that. I just want to know basic fundamental like what people teach kids in schools
Click to expand...
You want to learn SAS. Learning other languages while good from an academic sense will take more time.

How fast do you need to learn SAS?
 
blr_p said:
Get a tutor for SAS. That's what you want to learn.

You want to learn SAS. Learning other languages while good from an academic sense will take more time.

How fast do you need to learn SAS?
Click to expand...
In a year or so but SAS is only for Job purpose but I'm really interested in other languages like javascript and python
blr_p said:
Get a tutor for SAS. That's what you want to learn.
Click to expand...
But it's pretty expensive to learn though
 
chungus said:
In a year or so but SAS is only for Job purpose but I'm really interested in other languages like javascript and python
Click to expand...
can you give some examples of what exactly goes over your head? may peeps here including me can guide you better if we knew what exactly you arent getting, chances are you havent even started basic courses and are directly jumping in to some advanced topics
 
chungus said:
But it's pretty expensive to learn though
Click to expand...
See if you can find a tutor that can customise what you need to know. Aren't there online portals where people do this in India?

This is the fastest way to learn hence the cost and stay motivated. No different to a gym instructor
 
calvin1719 said:
If you want anything approaching accurate knowledge, do NOT ask chat bots to explain or tell you stuff.

I dunno how you think that's appropriate, vector, to advise a beginner to do that, given how much of a complete shitshow chatgpt's knowledge is at the moment.
Click to expand...
I am not in a technical role but for the lulz I completed various HackerRank challenges using Copilot and was on top of the organisation leader board for a while. I would say 80% of use cases can be covered by foundational models as-is and it will only get better with smaller, fine-tuned models in upcoming months.

Sorry to say, but the comments and debugging solutions provide by LLMs work in most cases. Also, lot of the data science processes are now automated with SaaS like Azure AI and it will only get better.
 
altair21 said:
can you give some examples of what exactly goes over your head? may peeps here including me can guide you better if we knew what exactly you arent getting, chances are you havent even started basic courses and are directly jumping in to some advanced topics
Click to expand...
Like loops, strings, full loops along the way the bare basic concepts
 
chungus said:
Like loops, strings, full loops along the way the bare basic concepts
Click to expand...
start from the bare basics then, look up a few basic python courses on reddit and go from there, the reason I'm recommending python to you is because it has the least cognitive load syntax wise, its straight forward af and will help you build up the concepts you can apply in other languages. I literally started my SIL from how to print hello world, and stuff like a=10, print(a), before she built up enough basics, for context, she's an analyst and a wizard in Excel, can probably do better than 90% of the data analysts I have seen, but wanted to change streams to get more remote-friendly jobs, granted she already had some basics built up but its just a matter of relating logic which you probably have been applying for years and matching it to programming
 
Back
Top