I started watching Alex in Jan 22, got my first data job in June, and just got offered a role as a data analyst this month. Increased my salary 60% in less than a year. Thank you Alex!
@@Pcswny I was working at a university monitoring attendance, it lent itself nicely to building my skills in excel, making charts, and performing my own analysis of the data. I did the Google cert as well which was really great, but I think applying skills to your current role where you can is really important so you have some measurable experience and can compete. So many people are doing the Google/IBM cert and you need something to help you stand out since the barrier to entry is pretty low.
Thank you for sharing. Im a 44 year old commercial fisherman trying to switch careers. I love the Google cert class so far and I can see how important this role is for the success of my business and for the success of any business!
Junior and even mid-level engineers will really benefit from simply being able to paste code and ask why the code was written in a particular way. I've given it some pretty complex SQL code and it explains it correctly.
@@mrs8551 I was thinking the same thing! The more I use it and other AI tools, the more I wonder. Present/near future I think it will enhance efficiency/output tremendously while pushing the envelope in terms of quality, creativity and overall impact. Long term, I could see those jobs evolve alongside AI, but anything could happen.
A great book on this is "Humans Need not Apply". It was written a few years ago but reading it in the context of this new AI is incredible. I am currently in a Masters program and you can imagine the impact it has had this semester.
At time 2:18 "I don't really believe in paying." I can understand that. But, you must believe in being paid given the join button. I believe in both. At any rate, thanks for the video. Great quality.
my wife laughs at me when I tell Alexa Please and Thanks. I'm just saying that I'll have a better time of it when the AI-Robot overlords take over cause they'll remember my kindness.
This is amazing stuff, Alex. I’ve used ChatGPT to translate calculated fields in Google Data Studio to Tableau and even to Power BI. It’s truly fascinating how it works! I think it’s going to be a total game changer. Personally, I think data engineers are at more risk of job loss from ChatGPT integration into our data analytics tools than data analysts. Can’t wait to see your next video on this though!
Lol everyone is at risk in the early years then we all transition to the new skills required. If I had to guess you building your coding skills will only benefit you. I think Data Architects are the ones who will actually benefit the most because they know how to piece the puzzle together and don’t need to be a full stack engineer
You were so polite and valued its work. I wonder how people growing up now, getting used to ai doing hard work for them even if they are absolutely rude, will affect our society.
Alex, I've gotta say... I've been thinking about switching careers for a while now and Data Analytics is right up my alley. I've watched some of your videos and your content has been nothing short of amazing and encouraging. I'm definitely switching this year, and 2023 will be my last year as a teacher! I'll let you know how it goes
always so inspiring Alex! I've used the explaining, generating, commenting approach myself and funny story, it actually gave me the wrong code once, and when I entered "can you correct the code because your code is giving me the wrong output for n=90" it corrected itself and gave me an optimized accurate code, so we def need to be careful on not 'what all we can do through chatgpt' but also 'how' to do it :)
I spent 35 yrs working in big data analytics at one of the credit bureaus - started out on mini computers and moved to the very first pc's so I have literally seen it all. I spent the last 3 months playing with GTP4 on a couple of private big data projects and in conjunction with an ETL tool (like alteryx) I can do in half a day what it took a small team of phd stats 3 months just 15 yrs ago. GPT is extremely impressive in generating code but equally impressive in suggesting methods. Without doubt GPT and it like are going to radically change data analysis and probably allow the more creative folks to win out over the less creative. It will mean fewer programmers and data engineers for sure. If I were 22 again I would make sure I had the widest possible understanding of business processes / problems and a wide range of analytical methods at a theoretical level. AI will plug the gap.
Thank you for your wisdom. I'm 23 and I feel like you wrote this for me. I was born an entrepreneur and I feel like I was born in the perfect age to take advantage of emerging technologies and solve rudimentary problems for business in my country. I am currently using GPT4 and a tool called RATH by Kanaries to study the historical sales data of my father's car spare parts shop. My dream is to turn the business into a luxury car customization shop. For now I want to study the data as much as I can to help the company make the most informed business decisions.
I believe ChatGPT's operating memory space, today, is 4097 (input+output) tokens, which approximates 3000 words. This is perhaps why you are seeing truncated outputs, or limited memory of historical input.
FYI, if the response from ChatGPT is cut out, its likely because the token for the conversation is run out. You can just prompt with "continue" or "go on" to continue writing previous response.
Just a tip when chat gpt stops like that you can just tell it to “continue your response from where you left , you stopped mid way “ and he will continue the previous response , pretty cool
Hi Alex, amazing video, I’ve always been interested in the data science field, and working full time as a data analyst. But I have a few questions about the same. It would mean so much to me if I could have just 10 minutes of your time and discuss those questions. And again, thanks for the great content 😊
Pretty much aligned with not just how but also why I use gpt. In a relatively short space of time, I replaced what I used to do with SQL and Excel with Python generated with gpt (no prior python or coding experience). Just like you showed in this example, just that in the end, I made Python files and global functions to run what I ran with the first few data sets I explored. Getting this phase of ETL so quickly implemented, brings the more fun and brain taxing part of working with the resulting data up front in no time at all. Which brings me space and time to keep going with gpt to implement more code to do tasks I need done, formulating them dynamically as I go so they can be reused, which leads me to think that a web based browser is on the horizon, to run my functions based on user inputs instead of me via an IDE. Which of course, one can use gpt for again to try out, build, refactor. In effect, I am building a tool for myself to replace a lot of what I usually find I need to do, just to get me somewhere where the real work of solution building begins, but with a tool box of data exploded from source data. And it seems no sign at all ahead, of what gpt and python can't help me with. It just today advised that for the next step of my actions (formulating structured data points from unstructured comment fields) that with Python I can utilize X, Y, Z libraries in A, B, C steps to approach my task. What is stunning is that you can ask gpt to do a task for you, like write some sql code - but you can then ask it to give you the python code it used to output the sql code to me, so that I can make a function and not have to come back to gpt again for the same ask. I don't give gpt any data - just tasks and examples. It would be fantastic to have a gpt on my local machine, with access to UI commands like I do, through keyboard and mouse referencing - to action these actions, to save files, write files, run queries... I don't really like the idea that any of my true code and data and use case, is sent to OpenAI however - for sure they must have an gpt like model reviewing all user inputs and gpt outputs and clustering on content and use per user interaction with gpt, anything we share is somehow also in their awareness and so customer data, is a no no to use on the platform but also I am concerned how long we will have free access to what we currently have access to. OpenAI know what I use gpt for, whats to say it sees that so many of us are doing similar that they don't pursue the +plus model for coding specific and thats then accessible only through paying... Its a fantastic tool, just the future looks very foggy to me as to where and how this will go.
The most interesting thing I would use AI for is to simulate time lines with the purchasing power of people. For example, buy a home at varying prices, time lines, and then measuring the quality of life on average following the guidelines. For example, make a shopping list of everything you ever want to buy in the next 30 years, then ask AI what prices everyone would have to sell those things under which timelines, then perhaps we could redesign economies more efficiently.
Awesome video Alex - thanks! When dealing with a much bigger dataset that is too many tokens for a single prompt; how do you propose using GPT to analyse the data? It looks like some people are using GPT in a Jupyter notebook to analyse the full dataset. I'm not an engineer so do not know exactly how to set this up. Would you consider showing how to do this in a future video?
Hey Alex. Thanks for great video as always! What do you think wich professions this technology ( or other AI) will change in near future? And what skill will be always in demand?
I was trying chatgpt to help optimizing my SQL codes & didn't realize it can take the entire dataset as an input. When you gave example by putting the dataset & asked chatgpt to provide insight plus recommendations, that scared the hell out of me. I just got laid off at 43, & was trying to switch career to data analyst by looking at your videos & taking local language course to speed up. Then this chatgpt came.....it's really a bad timing hehehe
Information without context is a dangerous thing. You still need to add the addtional context to the chat gpt high level insights to make the findings actionable. saying focus on low selling products, thats stock standard 101, but which products? how many are there? there are multiple layers to this space
I made a gui that allows me to open a .txt file and convert it to a .csv and you can select your file path to save the new .csv file the possibilities for this are amazing!
It's okay. I think it's a good thing when you're polite to the AI I mean... Maybe in the future future AI is going to see how we treated the originals....
Guys, to be honest, you should not much rely on GPT3. Here is what it told me: ME: Is it possible to get the wrong code from your side? GPT3: As an AI language model, I don't have the ability to deliberately provide incorrect code. However, it's always possible that there could be mistakes or typos in the code that I provide, so it's a good idea to double-check the code and results to ensure they are correct.
Where your dataset stopped, it is hitting the size limit (it is trying to pass your whole chat to the gpt api) or on the free one if servers are too busy. Most of the time you can just type continue, and it carry on from where it stopped. when I found that keyword it made my life a lot easier. (on GPT4, not sure if it is available on the paid version yet. I think it is only for approved beta users, it can go up to 32K in size, gtp3.5 turbo can go up to 4k, and i think gpt3 is 2k)
Hi , I encounter the same issue when I try to format all my data in chatgpt. And typing "continue" still doesn't enough to work through the rest of the data. Do you have any advice on this ?
if you tell chat-gpt what version of the programming language you are using it will give you better answers, otherwise it is trying to figure it out through context clues either in your code or from your conversation
Its not the server overloading why its erroring out. It has a max of 4000 tokens (± 3000 words) that's why it stops. Just insert your data before and ask it to continue from where it stopped and it'll finish your code for you
Often times when it stops outputting, you can simply type "Continue" as your next prompt. The other thing you should know is that it doesn't stop because the server is overloaded, but because you've reached a character limit in the output. For the same reason you cannot do real data analysis with ChatGPG. Try adding a dataset of greater than like 100 rows and 10 columns. It won't work. You can break out the data into pieces and feed it prompt by prompt, but then you run into a data memory retention issue. You shouldn't tell people they can do analysis with ChatGPT unless their dataset is miniscule. It won't work otherwise
you don't need to put 1M lines of data in chatgpt, the idea is that just from data header get you some ideas of analysis and write e.g. python code which run all the data at google colab for example. btw for ideas of analysis one should not ask about pure table, but to explain the context of the data, and meaning of vars in column definitely. Then you will get more usefull answer.
@@purgenetik then you can only ask to give some ideas of what type of analysis you can do, but not to analyze your dataset or to do data analysis like he is showing on this video. But anyway now with the new Codex plugin you can actually give it a large CSV file and it will do what this video is showing, which is full data analysis of your dataset.
For people watching this before starting their data analysis journey - NO this will not make much sense. Yes, learning basics of of data analysis, python, SQL, etc. is required to know the jargons/language of how to word your prompts. Alex has provided some certifications in the desription to get started.
You seem to be more excited with the technology than the results. I have been trying to accomplish one real life common, simple, fiduciary problem. I ran hundreds of variations to get the correct answer.. ChatGPT has yet to come close to solving the problem that I can just look at and solve.. I would have no problem paying for the product, if it worked.
Hey Alex. Currently watching your SQL bootcamp right now, and also waiting for this video. What environment would you preferably recommend between MySQL and PostgreSQL? THANK YOU.
yeah i would like to know more about this too. From what i can tell MySQL is used more in industry, but PostgreSQL is the most loved by its actual users.
While this is useful in a stand alone context, the most benefit will come when it is fully integrated into coding environments and data analysis tools with training optimized for those specific activities. Right now just using chatgpt in a stand alone context is limited and for full use of this functionality you still need data engineers, programmers and so forth to build applications around it. One problem, however, is the cloud nature of chatgpt and the fact that they ultimately want you to pay for it, which limits how it can be used in open source projects.
ChatGPT has a symbol limitation. When it stops for no reason, this is the reason. Ask it to continue and you will have the final result in a couple of dialogs.
I agree with you that it's amazingly impressive. They're not really using code to do this. They're using a massive neural net trained on petabytes of language data, and a bit of feedback on what we humans perceive as helpful or not.
I'm asking this on several channels, brand new to ChatGPT and have been trying to sign up since Monday (this is Wednesday, so 2 days now). I always receive the same error message saying sign-up is not available. Any suggestions?
Anyone can tell me if that would be the best setup to analyze Roulette spins made by a specific table? Also I'm trying to build the perfect prompt to get GPT to tell me the next number (within 6 chips) using that info, like an algorithm would. I've seen a few built that gives the exact number, but they won't share or sell.
All examples are for small datasets of a few records. Is it possible to use chat GPT for analysis with a larger dataset of say 500 or 800 records (40 variables); how?
they dont need any data anylyst, data engineers, data scientists, software developers, software engineers and basically anything in Tech. But there will still be here. In the near future AI will automate all tech jobs though.
@@vectoralphaSec how if my true passion in life is to buy 5-10 different mansions in the most expensive cities in the world, and party with 20 beautiful girls every day? will UBI allow me to do that?
In machine learning (ML) the dichotomy developers face is providing accuracy at the expense of precision and vice versa. The issue is that you can't maximize both in a ML model so you have to settle for an acceptable balance of the two which varies from application to application. And this is the beauty of the human mind. I don't personally think that humans will ever be able to completely play God and produce a replica of the human mind (in terms of programmatically creating things like intuition, instinct, and to the greatest extent, contextual decision making). So in some ways, yes, I believe it could replace basic analysis and even some complex analysis. But it'll still need a human to provide it with the necessary context to give it the parameters it needs to produce accurate results.
I started watching Alex in Jan 22, got my first data job in June, and just got offered a role as a data analyst this month. Increased my salary 60% in less than a year. Thank you Alex!
What did you do before?
@@Pcswny I was working at a university monitoring attendance, it lent itself nicely to building my skills in excel, making charts, and performing my own analysis of the data. I did the Google cert as well which was really great, but I think applying skills to your current role where you can is really important so you have some measurable experience and can compete. So many people are doing the Google/IBM cert and you need something to help you stand out since the barrier to entry is pretty low.
@@harrietdrumswhat kind of job did you get at first and how did you prepared for it ?
I have to jump in. Alex helped me land a position last year at a wonderful fintech company. Thank you Alex.
Thank you for sharing. Im a 44 year old commercial fisherman trying to switch careers. I love the Google cert class so far and I can see how important this role is for the success of my business and for the success of any business!
Junior and even mid-level engineers will really benefit from simply being able to paste code and ask why the code was written in a particular way. I've given it some pretty complex SQL code and it explains it correctly.
So ready…this is going to take data analysis and visualization to a whole new level!
Honestly it takes ALL of programming to a whole new level.
Will chatgpt kill data analyst,data scientist, and programmer job?.
@@mrs8551 I was thinking the same thing! The more I use it and other AI tools, the more I wonder. Present/near future I think it will enhance efficiency/output tremendously while pushing the envelope in terms of quality, creativity and overall impact. Long term, I could see those jobs evolve alongside AI, but anything could happen.
Gonna take our jobs literally
A great book on this is "Humans Need not Apply". It was written a few years ago but reading it in the context of this new AI is incredible. I am currently in a Masters program and you can imagine the impact it has had this semester.
At time 2:18 "I don't really believe in paying." I can understand that. But, you must believe in being paid given the join button. I believe in both. At any rate, thanks for the video. Great quality.
"I don't really believe in paying for things" is a statement that he should probably explain.
my wife laughs at me when I tell Alexa Please and Thanks. I'm just saying that I'll have a better time of it when the AI-Robot overlords take over cause they'll remember my kindness.
😂😂
Hi Alex, I am binge watching your Bootcamp since the day it's launched. Thank you so much for your efforts. Wish me luck.
Greetings all the way from Uganda Kampala, this channel is so inspiring. Thanks Alex for being a such a good person, you're really changing our lives.
This is amazing stuff, Alex. I’ve used ChatGPT to translate calculated fields in Google Data Studio to Tableau and even to Power BI. It’s truly fascinating how it works! I think it’s going to be a total game changer. Personally, I think data engineers are at more risk of job loss from ChatGPT integration into our data analytics tools than data analysts. Can’t wait to see your next video on this though!
Lol everyone is at risk in the early years then we all transition to the new skills required. If I had to guess you building your coding skills will only benefit you. I think Data Architects are the ones who will actually benefit the most because they know how to piece the puzzle together and don’t need to be a full stack engineer
@@JP-sm1dc exactly this and hence im really excited
BI Engineers may be the most at risk. All they doing is writi g SQL
Very nice use on Google Data Studio as I find the functions to have very strange formatting. It's easy to flub.
You were so polite and valued its work. I wonder how people growing up now, getting used to ai doing hard work for them even if they are absolutely rude, will affect our society.
Thanks!
Thanks for supporting the channel! :D
Alex, I've gotta say... I've been thinking about switching careers for a while now and Data Analytics is right up my alley. I've watched some of your videos and your content has been nothing short of amazing and encouraging. I'm definitely switching this year, and 2023 will be my last year as a teacher! I'll let you know how it goes
We are in the same page ... Can not quite the teaching job and switch to Data job
Good luck to you.
I’m in the same boat. This AI is fascinating.
On the same page.Can we form a partner accountability group...
@mariem151 I would like to join the group too
hows it going bro ?
always so inspiring Alex! I've used the explaining, generating, commenting approach myself and funny story, it actually gave me the wrong code once, and when I entered "can you correct the code because your code is giving me the wrong output for n=90" it corrected itself and gave me an optimized accurate code, so we def need to be careful on not 'what all we can do through chatgpt' but also 'how' to do it :)
That's super true - happens all the time
I was waiting for this video. Looking forward to hearing your thoughts. Thanks Alex!!
I think it's extremely fascinating and really useful! Hope you enjoy it! :D
Nice! This is the video I have been waiting for 😊
I spent 35 yrs working in big data analytics at one of the credit bureaus - started out on mini computers and moved to the very first pc's so I have literally seen it all. I spent the last 3 months playing with GTP4 on a couple of private big data projects and in conjunction with an ETL tool (like alteryx) I can do in half a day what it took a small team of phd stats 3 months just 15 yrs ago. GPT is extremely impressive in generating code but equally impressive in suggesting methods. Without doubt GPT and it like are going to radically change data analysis and probably allow the more creative folks to win out over the less creative. It will mean fewer programmers and data engineers for sure. If I were 22 again I would make sure I had the widest possible understanding of business processes / problems and a wide range of analytical methods at a theoretical level. AI will plug the gap.
Thank you for your wisdom. I'm 23 and I feel like you wrote this for me.
I was born an entrepreneur and I feel like I was born in the perfect age to take advantage of emerging technologies and solve rudimentary problems for business in my country.
I am currently using GPT4 and a tool called RATH by Kanaries to study the historical sales data of my father's car spare parts shop. My dream is to turn the business into a luxury car customization shop. For now I want to study the data as much as I can to help the company make the most informed business decisions.
I believe ChatGPT's operating memory space, today, is 4097 (input+output) tokens, which approximates 3000 words. This is perhaps why you are seeing truncated outputs, or limited memory of historical input.
FYI, if the response from ChatGPT is cut out, its likely because the token for the conversation is run out. You can just prompt with "continue" or "go on" to continue writing previous response.
makasi ngab
Thank you so much for this information. Have recently just started trying out chatgpt and had the problem and didn't know what to do.
Just started using Chat GPT and wondered when could I expect a vid from Alex the Analyst.
Just a tip when chat gpt stops like that you can just tell it to “continue your response from where you left , you stopped mid way “ and he will continue the previous response , pretty cool
Oh nice! I'll try that!
It* not he
wow Thanks for that gem Pablo, I went back and used it on a few of my older chats where is suddenly stopped.
Like the intitial relase of Power BI in 2014, this is going to be a game changer for sure!
Love this channel
Hi Alex, amazing video, I’ve always been interested in the data science field, and working full time as a data analyst. But I have a few questions about the same. It would mean so much to me if I could have just 10 minutes of your time and discuss those questions.
And again, thanks for the great content 😊
I’ve started learning the basics of Python and SQL
wow Alex, tremendous video! Thanks for sharing! I'm blown away!
(By the way, you can ask “Did you time out?” and it will retry the last answer, often completing it instead of stalling like it did previously.)
Loved the video! Please make more on how to use AI like this. 🤩
Will do!
Excited to see this!
Hope you can share on how will this impact on data analyst industry? Will it replace the job or enhance the industry?
Pretty much aligned with not just how but also why I use gpt. In a relatively short space of time, I replaced what I used to do with SQL and Excel with Python generated with gpt (no prior python or coding experience). Just like you showed in this example, just that in the end, I made Python files and global functions to run what I ran with the first few data sets I explored. Getting this phase of ETL so quickly implemented, brings the more fun and brain taxing part of working with the resulting data up front in no time at all. Which brings me space and time to keep going with gpt to implement more code to do tasks I need done, formulating them dynamically as I go so they can be reused, which leads me to think that a web based browser is on the horizon, to run my functions based on user inputs instead of me via an IDE.
Which of course, one can use gpt for again to try out, build, refactor. In effect, I am building a tool for myself to replace a lot of what I usually find I need to do, just to get me somewhere where the real work of solution building begins, but with a tool box of data exploded from source data. And it seems no sign at all ahead, of what gpt and python can't help me with.
It just today advised that for the next step of my actions (formulating structured data points from unstructured comment fields) that with Python I can utilize X, Y, Z libraries in A, B, C steps to approach my task. What is stunning is that you can ask gpt to do a task for you, like write some sql code - but you can then ask it to give you the python code it used to output the sql code to me, so that I can make a function and not have to come back to gpt again for the same ask.
I don't give gpt any data - just tasks and examples.
It would be fantastic to have a gpt on my local machine, with access to UI commands like I do, through keyboard and mouse referencing - to action these actions, to save files, write files, run queries... I don't really like the idea that any of my true code and data and use case, is sent to OpenAI however - for sure they must have an gpt like model reviewing all user inputs and gpt outputs and clustering on content and use per user interaction with gpt, anything we share is somehow also in their awareness and so customer data, is a no no to use on the platform but also I am concerned how long we will have free access to what we currently have access to. OpenAI know what I use gpt for, whats to say it sees that so many of us are doing similar that they don't pursue the +plus model for coding specific and thats then accessible only through paying... Its a fantastic tool, just the future looks very foggy to me as to where and how this will go.
Thanks for the detail use cases of ChatGPT in programming.
I'm eagerly waiting for this video.
You should give the Noteable plugin a try, it can do some pretty amazing things.
This is impressive ✌🎖
Thank you for another interesting and relevant video for data analysis!
The most interesting thing I would use AI for is to simulate time lines with the purchasing power of people. For example, buy a home at varying prices, time lines, and then measuring the quality of life on average following the guidelines.
For example, make a shopping list of everything you ever want to buy in the next 30 years, then ask AI what prices everyone would have to sell those things under which timelines, then perhaps we could redesign economies more efficiently.
It will be interesting to try these in GPT-4 now. 😃
Alex,you are heaven sent
Awesome video Alex - thanks! When dealing with a much bigger dataset that is too many tokens for a single prompt; how do you propose using GPT to analyse the data? It looks like some people are using GPT in a Jupyter notebook to analyse the full dataset.
I'm not an engineer so do not know exactly how to set this up. Would you consider showing how to do this in a future video?
Hey Alex. Thanks for great video as always! What do you think wich professions this technology ( or other AI) will change in near future? And what skill will be always in demand?
Another excellent video. Thanks.
I was trying chatgpt to help optimizing my SQL codes & didn't realize it can take the entire dataset as an input. When you gave example by putting the dataset & asked chatgpt to provide insight plus recommendations, that scared the hell out of me. I just got laid off at 43, & was trying to switch career to data analyst by looking at your videos & taking local language course to speed up.
Then this chatgpt came.....it's really a bad timing hehehe
Information without context is a dangerous thing. You still need to add the addtional context to the chat gpt high level insights to make the findings actionable. saying focus on low selling products, thats stock standard 101, but which products? how many are there? there are multiple layers to this space
Thank you so much!
impressive, thank you
I made a gui that allows me to open a .txt file and convert it to a .csv and you can select your file path to save the new .csv file the possibilities for this are amazing!
It's okay. I think it's a good thing when you're polite to the AI I mean... Maybe in the future future AI is going to see how we treated the originals....
Waiting to watch it.
👋 Alex is your data analyst bootcamp Playlist enough to getting job for beginners..??😊😊
I'm waiting to watch this video.
Great video. I'm curious to know though. What do you mean by "you don't believe in paying"?
It was mostly a joke - I don't like paying for things usually lol I will if it's worth it though
Guys, to be honest, you should not much rely on GPT3. Here is what it told me:
ME: Is it possible to get the wrong code from your side?
GPT3: As an AI language model, I don't have the ability to deliberately provide incorrect code. However, it's always possible that there could be mistakes or typos in the code that I provide, so it's a good idea to double-check the code and results to ensure they are correct.
This is why stack exchange has banned chat gpt answers for the time being.
Where your dataset stopped, it is hitting the size limit (it is trying to pass your whole chat to the gpt api) or on the free one if servers are too busy. Most of the time you can just type continue, and it carry on from where it stopped. when I found that keyword it made my life a lot easier. (on GPT4, not sure if it is available on the paid version yet. I think it is only for approved beta users, it can go up to 32K in size, gtp3.5 turbo can go up to 4k, and i think gpt3 is 2k)
Hi , I encounter the same issue when I try to format all my data in chatgpt. And typing "continue" still doesn't enough to work through the rest of the data. Do you have any advice on this ?
When it gets stuck, you can just ask to "continue" and it will continue where it left off.
You can ask it to continue when it gives you just a part of a code and it will
if you tell chat-gpt what version of the programming language you are using it will give you better answers, otherwise it is trying to figure it out through context clues either in your code or from your conversation
Its not the server overloading why its erroring out. It has a max of 4000 tokens (± 3000 words) that's why it stops. Just insert your data before and ask it to continue from where it stopped and it'll finish your code for you
Hi Alex,,,Do you think ChatGPT will make the role of a Data Analyst redundant?
Often times when it stops outputting, you can simply type "Continue" as your next prompt. The other thing you should know is that it doesn't stop because the server is overloaded, but because you've reached a character limit in the output. For the same reason you cannot do real data analysis with ChatGPG. Try adding a dataset of greater than like 100 rows and 10 columns. It won't work. You can break out the data into pieces and feed it prompt by prompt, but then you run into a data memory retention issue. You shouldn't tell people they can do analysis with ChatGPT unless their dataset is miniscule. It won't work otherwise
you don't need to put 1M lines of data in chatgpt, the idea is that just from data header get you some ideas of analysis and write e.g. python code which run all the data at google colab for example. btw for ideas of analysis one should not ask about pure table, but to explain the context of the data, and meaning of vars in column definitely. Then you will get more usefull answer.
@@purgenetik then you can only ask to give some ideas of what type of analysis you can do, but not to analyze your dataset or to do data analysis like he is showing on this video. But anyway now with the new Codex plugin you can actually give it a large CSV file and it will do what this video is showing, which is full data analysis of your dataset.
Thanks for a great video.
How can I analyze big data sets with chatgpt or any other AI engines?
Right now we are waiting on some larger ai systems like in Microsoft Azure, but they will get there soon
@@AlexTheAnalyst Thanks
just type "continue" and it will
19:00
I am not new to data analysis, but am new to actually doing it systematically, Should I start elsewhere in the series? I am very familiar with gpt.
AWESOME!!!
Some of us may have to pivot real quick from BA/DA to something more accommodating..
Have you looked into the token use you get with each chat? Once you hit the limit you need to start a fresh chat
Ottah Daniel from Nigeria. Taking Google data analytics professional course Coursera
Hi Alex. Can you work with chat-gpt to find relationship between data to do some predictive analysis?
Very interesting
For people watching this before starting their data analysis journey - NO this will not make much sense. Yes, learning basics of of data analysis, python, SQL, etc. is required to know the jargons/language of how to word your prompts. Alex has provided some certifications in the desription to get started.
Alex I seriously miss your portfolio series when are you going to start with series looking forward
It's a mind-blowing things
How do you save the different chats that you have on the left into tabs or folders?
I can’t wait anymore
Is this going to take over the data analyst job?
You seem to be more excited with the technology than the results. I have been trying to accomplish one real life common, simple, fiduciary problem. I ran hundreds of variations to get the correct answer.. ChatGPT has yet to come close to solving the problem that I can just look at and solve..
I would have no problem paying for the product, if it worked.
So, is it pointless to get a Bachelors in Data Analytics?
Hey Alex.
Currently watching your SQL bootcamp right now, and also waiting for this video.
What environment would you preferably recommend between MySQL and PostgreSQL?
THANK YOU.
yeah i would like to know more about this too. From what i can tell MySQL is used more in industry, but PostgreSQL is the most loved by its actual users.
Thanks Alex
I'd love to see a more full analysis example.
cool, I use ChatGPT for Sql code, most time give 100% correct code,
While this is useful in a stand alone context, the most benefit will come when it is fully integrated into coding environments and data analysis tools with training optimized for those specific activities. Right now just using chatgpt in a stand alone context is limited and for full use of this functionality you still need data engineers, programmers and so forth to build applications around it. One problem, however, is the cloud nature of chatgpt and the fact that they ultimately want you to pay for it, which limits how it can be used in open source projects.
Ironic when you consider it's developed by "open" AI lol
How would it work for SEO?
Hi Alex, you welcome
Can you performa ANOVA using chatgpt prompts?
Awesome
You need to do a video explaining WHY this isn't valuable enough to pay for!
When the answer stops in the middle, type "continue" and press return. If it was not too late, it will continue where it stopped
I have been using this non stop lol
ChatGPT has a symbol limitation. When it stops for no reason, this is the reason. Ask it to continue and you will have the final result in a couple of dialogs.
I agree with you that it's amazingly impressive. They're not really using code to do this. They're using a massive neural net trained on petabytes of language data, and a bit of feedback on what we humans perceive as helpful or not.
I'm asking this on several channels, brand new to ChatGPT and have been trying to sign up since Monday (this is Wednesday, so 2 days now). I always receive the same error message saying sign-up is not available. Any suggestions?
noob question here but if its already possible what about in 3 years ? will data analysts jobs all obscelete ?? thanks !
I'll have a whole video on this next week :)
@@AlexTheAnalyst nice I hope this is good news haha
Anyone can tell me if that would be the best setup to analyze Roulette spins made by a specific table? Also I'm trying to build the perfect prompt to get GPT to tell me the next number (within 6 chips) using that info, like an algorithm would. I've seen a few built that gives the exact number, but they won't share or sell.
Yes I made it!!!
For ChatGPT, profit equals revenue?
All examples are for small datasets of a few records. Is it possible to use chat GPT for analysis with a larger dataset of say 500 or 800 records (40 variables); how?
Not at the moment, but very soon it will be
00:02:18 "I don't really believe in paying". Huh?
so... with ChatGPT they dont need Data Analysts anymore???
they dont need any data anylyst, data engineers, data scientists, software developers, software engineers and basically anything in Tech. But there will still be here. In the near future AI will automate all tech jobs though.
@@avellian6840 UBI and they follow what their true passion in life is without worrying about money.
@@vectoralphaSec how if my true passion in life is to buy 5-10 different mansions in the most expensive cities in the world, and party with 20 beautiful girls every day? will UBI allow me to do that?
DO YOU THINK CHAT GPT CAN REPLACE THE JOB OF DATA ANALYST ? PLEASE REPLY...
In machine learning (ML) the dichotomy developers face is providing accuracy at the expense of precision and vice versa. The issue is that you can't maximize both in a ML model so you have to settle for an acceptable balance of the two which varies from application to application.
And this is the beauty of the human mind. I don't personally think that humans will ever be able to completely play God and produce a replica of the human mind (in terms of programmatically creating things like intuition, instinct, and to the greatest extent, contextual decision making). So in some ways, yes, I believe it could replace basic analysis and even some complex analysis. But it'll still need a human to provide it with the necessary context to give it the parameters it needs to produce accurate results.
Wowowowowow
(please don't take this seriously) can you do analysis and visualization of how many hairs present or will be present their on your head..
I have applied for a chatGPT account but I am on the waiting list