Go from Excel novice to data analysis ninja in just 2 hours: kevinstratvert.thinkific.com/ Get the latest high-quality tutorial and tips and tricks videos emailed to your inbox each week: kevinstratvert.com/newsletter/
That is the business model of OpenAI and other upcoming AI companies. They will offer businesses customized, closed and secure AI engines that could only be used within the confines of the company's network.
Excel VBA is still the most widely use tool for engineering and finance. Wait until you have a question from an external auditor. You still need to know the underlying process. Engineer questions are not that simple to infer. For example I just asked co pilot why the coefficient of correlation doesn’t display correctly in charts sometimes and he didn’t know why.
Not excel, manual or formula data analysis is dead already. Excel is a data structure in tabular form to manipulate. All the power of analysis is on the AI side soon which will be faster and trivially done by anyone without knowing Excel. You can store all the data you want in it and just cause it's a table, it will survive as a data saving format. But not for much longer.
I understand both points of view, new technologies have been trying to kill excel for more than 20 years now unsuccessfuly... But maybe this time is different? But also notice that when asked why profits when down the last year, the AI suggested different reasons without providing the correct answer... This will mean that the data analyst will still need to have skills in excel, powerbi or other data analyst tools to drill in and find the currect reasons... I believe that the data analysts that will survive the AI apocalypse are the one that can automate and speed up their work working WITH this new AI tools, no against them... We have done it before. When powerbi appeared, people that adapted and learned and added that tool to our arsenal did very well... AI could be another tool that data analyst power users can use to become faster and more skilled in our job... Beside, you still need to ask the right questions to the AI and decide on the best provided solutions, and that is exactly the kind of things that good data analysts can do very well...
Excel has huge problems. You cannot use it in git. You cannot test it in a god way. The amount of data is limited and its slow. The reason why its still in use is, because people with an economics background need bullshit jobs. Thats why ai systems won't kill Excel
If you think this UA-cam video title is an accurate representation of what the creator thinks or what the video is about and not just a clickbait title then you don’t know UA-cam
This is fantastic for ease of use, but slightly terrifying as an analyst watching a tool created that can allow entry-level hires to complete high-level tasks with no actual knowledge (which were prerequisites for hire initially)
I suppose my biggest solace lies in the fact that I’m employed in a secure facility. The data we work with requires security clearance to access it, so I don’t need to stress about losing my job to some yahoo who used this tool to “look good without learning” since we would never be permitted to have Chat-GPT on our devices or export a file to use on a personal device.
@@PolyInProgit could change drastically there are LLMs which are being developed which will run on premise servers , training will be done on Large Datasets but no data will be sent outside 😂 no one is safe in AI revolution
It can allow entry-level people to make huge and horrifying mistakes because they can't tell what they are looking at, just like highly automated industrial equipment can be run by a noob, but...
Chat GPT is a wonderful tool for somethings, but I for one would not be prepared to upload my very sensitive financial data to anywhere in the "cloud" to anyone for any reason!
As someone who is an analyst, this is both terrifying and amazing. However, your data is very clean. If you give it some data that’s dirty, it does a terrible job and will fail.
...this won't replace data analysts because you still need someone to curate and collect the data and make certain there's no discrepancies in the real world. If anything, it makes a Dat analysts life easier. You're always going to need a human to communicate. Experience is key.
AI will communicate it better than a human. I think you are overestimating humans in that respect. It's early days for AI and it moves and improves faster than we can imagine.
It might not replace human analysts per se but it will definitely have an impact on how many analysts it takes to do the same amount of work. If AI tools help reduce the need for technical know-how and allow analysts to do a piece of work 2x-3x as fast, that increases the pool the people who could do analysis AND reduces the number of analysts required, a double whammy when it comes to the job market for analysts.
@@distilledgogilbaThere is hell lot of big data to be analysed .. Excel in this era is totally outdated as it hardly handles a million row data .. You need skills to work on cloud, build ETL to automate process and connect to a BI tool which again comes with set of measures to be created, BI has to be optimised to perform stutter free especially on realtime big data.. plus data handling comes with set of security that needs to be handled.. First understand what data analyst do and then think how AI will help in certain tasks.. I use chat gpt to correct my DAX measure or M codes, create customised plots through python, understand business problems if it's out of my domain and many more.. The one showed in video is very basic for industry standards.. Don't get panicked or panic others who are passionate in this area..
Chat is really useful. It’s even more useful if you have some sort of background in what you’re trying to excel at (no pun intended). That’s why people are so scared to use it, it can be used for almost anything but most people don’t have the necessary information and knowledge to maximize its use.
This is true, but these functions are better exploited by analysts because they have the knowledge of how to perform analysis and what type of analysis is most appropriate for various cases.
Most of the time that happens I’ve hit a dead end. If you start over and summarize what you gained from the last conversation it gets going pretty well
Can confirm. I threw a sheet with different daily rates for 3 roles in 3 tiers each by staffing companies ( procurement data) . WIth the initial prompt "Which company is the cheapest" and a couple of replies to GPT asking me I arrived at a list cheapest to most expensive on average. WIld.
this is simply insane ! makes works so smarter and efficient and helps in free up time to pick up some more pressing tasks . thanks, Kevin, for sharing tips & tricks along with tutorials. You have touched many lives .
I completely agree. I believe the future of data analysis will not primarily hinge on proficiency with tools like Excel, SQL, Python, or R. Instead, it will revolve around effectively communicating with AI systems, such as ChatGPT, and other emerging technologies. Mastering AI language will become a crucial skill across all professions. However, a solid understanding of your specific professional functions will be essential. It will enable you to craft precise prompts, ask relevant questions, and identify potential errors in AI-generated responses.
After seeing this video, I can confidently say excel is alive and well. I work in excel for a living and I am probably one of its most proficient users.
A user can own Excel and can use it offline Nor are they required to upload/share their data to some 3rd party, chatGPT. Still very useful,, especially when working with data that isint sensitive
Great video! You missed to show that you can ask GPT for a downloadable png of the graphs You can also do so much more, like flowcharts, mind maps, heat maps, etc
@@KevinStratvert In my case, I had a sheet with several thousand rows of Amazon sales data, including daily and hourly data. GPT was able to analyze it and give me deep insights, such as hourly and daily trends. Plus, it provided me with super cool graphs to show YoY comparisons and hourly sales heat maps too. I was super impressed and actually found it very useful.
Thinking on this more, beyond the magic of seeing the GPT engine produce analysis, the Achilles heel of this tool is its single use and lack of memory. There is no memory after the session of questions. Every spreadsheet analysis is one off. Super the boss might think it’s amazing the first time, but as soon as the spreadsheet changes - and it always does - you’ll have to start the chat and all your questions and charts all over again. Suddenly it doesn’t seem such a time-saver. Meanwhile, if you get your data person or the Excel whiz 2 seats over to make a pivot table, chart, or a full data model and automated dashboard, they’ll update instantly as new data gets put in, changed, or corrected. Waaaay better for anything more than the most simple of questions any business not started literally yesterday should already have answers to.
I remember my excitement in 1999 when Google Search became open for the public at large. Haven't had that same feeling in 25 years until now with ChatGPT 😊
I think it's nice that AI is able to read through spreadsheets and derive some meaning, but it's still a far cry from actually working with the data. These tools are still evolving not able to handle complex or dirty data and have limits. It's the same as using AI for any other work, you must be aware of what it knows, ask your questions carefully and double check the results. At the moment the only danger to anyone's job will come from managers who don't understand the job. It's up to you (the data analyst) to make them aware of what you do over what the tool does. Turn it on its head and show how the tool could help you but not replace.
Yeah once it starts interpreting dirty data and cleaning it up like a good human analyst does but in 10 seconds vrs an hour for a human that will be cool to see
To be honest it looks good, but i would save time if i use directly a pivot table and insert a chart, at some point you need to do stuff by yourself and not always rely on AI.. :)
Exactly. This IS done better with software. And you can save that file too. You can’t save the output from Chat GPT because it doesn’t produce a saveable file.
Indeed. This can help some total novice, but nothing showed here looks like a time saver, or perhaps those verbal interpretations could be if I write bullet points/hypothesis and ask gpt to elaborate from those.
You could also upload screen shots of analysed data whether its tables, charts, or anything else and ask it to analyse it for you. It's a revolutionary learning and productivity machine and I love it 😎💪 That said, it might make mistakes sometimes. So, you should always be vigilant, careful, and even double check the information if you suspect something isn't accurate. It doesn't just do Excel but SPSS, JASP, R, Python and much more. Go have fun with it and happy coding 😁
Do not worry, it is ok to use AI for easy use, but when you have data with thousands of row and columns, even AI cannot accurately analyze that. Imagine even excel crash with those types of data and that is why you would need Power BI or Tableau. Second and maybe it should have been first is privacy of data. There are a lot of companies that don't want their data put in AI. And they clearly specify to not use AI.
Dude just gave all his company's secure information to bigdata and now all his competitors can have access to it for a price. Later, the Cookie Company went belly up.
If you upload an Excel file into ChatGPT as highlighted in this video, would you not worry that all the data within the file becomes part of the AI, and therefore part of the whole AI realm? If that's the case, then your data is part of the AI realm, and might be accessible to/by other users.
Can ChatGPT do cluster analyses or similar? That'd be really helpful. If it can, do the same data assumptions apply, or would it be able to recognize clusters without making sure these assumptions are met. Thanks, Kevin. You rock.
Chat GTP sounds good in theory, but when you need to load your own data into it, (using a spreadsheet) as how your supposed to avoid using (because you said it was too cumbersome, no one wants to use formulas) Where doss your info go, how it's used, who has access to it. Where is the privacy?
That last quote was a banger Kevin!!! “You don’t need to know how to use the tools , you just need to ask the right questions and to make the right decisions.”
In the case of a Data Analyst's employer - company data, that is sensative. Employers expect you to keep their data within thier intranet or approved vendor portals. You will not be able to drop any such files into this tool, sharing it with non company systems is a violation of InfoSec policy.
🎯 Key Takeaways for quick navigation: 00:00 *🤖 Introduction to ChatGPT Data Analyst* - Introduction to ChatGPT's functionality as a personal AI Data Analyst. - Overview of premium subscription features and custom GPTs. - Demonstrations of various GPT capabilities, including data analysis and visualization. 02:16 *📊 Data Analysis with ChatGPT* - ChatGPT's ability to analyze data from Excel files. - Demonstrations of simple and complex data analysis queries. - Explanation of how ChatGPT aggregates data from multiple sheets and performs calculations. 06:19 *💵 Data Visualization with ChatGPT* - ChatGPT's capability to create visualizations, such as bar charts, from data. - Demonstrations of modifying chart properties, such as axis formatting and bar colors. - Ease of making visual adjustments through simple text-based commands. 07:42 *📈 Advanced Data Analysis Queries* - ChatGPT's ability to provide detailed analysis and insights for complex queries. - Examples of querying trends, identifying potential reasons for flat profits, and analyzing seasonal patterns. - Illustration of ChatGPT's potential to assist in decision-making processes and business insights. Made with HARPA AI
If you don’t want to pay, just use Python and R… the codes for data analytics are done, you don’t have to think for the input. Only for the output. If there is something complex, it is the real life job, not how to code the analysis.
Will it coordinate an in depth analysis of internal and world event driven items such as the loss of a key employee, bringing on a new vendor, a flood in the region, the Suez Canal impacting just in time delivery dates, changes in laws or government, multiple data souces… and things you have not considered, when making predictions?
I gave ChatGPT 850 lines of data and when the results didn’t make sense she apologized and said she only used a sample of the data and then gave me the correct results using all data. The fact that I had to do that made me not trust ai right now.
as a former Microsoft employee! i now feel good I quit my job as well! I used PowerBI when PowerBI was a God in the industry! now what can PowerBI be with this AI capabilities
I’ve realised that people who don’t like AI starts their process like this. Stage 1: AI won’t be able to do XXX. Stage 2: AI won’t be able to do better than humans in XXX. Stage 3: WE SHOULD REGULATE THIS ASAP!!
Sumif equation for the revenue for one customer, reports though with tables and graphs definitely could hasten doing analytics on a per workbook basis.
Excel is far from being just an analytics tool. I've been using spreadsheets since the days of the old paper columnar pads, then Lotus 123, and then Excel. Excel and the others are often points of original entry for data. So, no, Excel isn't dead; it will be around for a while longer.
It’s not that important how chat GPT figures out how to make a poem or picture. But if you are relying on chat GPT to look through a bunch of spreadsheets and say - “your company is profitable” or “it’s losing money” I wouldn’t just say “Ok! I’ll go with that and make a presentation to the shareholders now”.unlike other things in AI you need to know how chat GPT is coming up with these answers.
But can it add data to the spreadsheet? I'd love to have it as an assistant that you could tell to add a new stock purchase to my existing portfolio in all the right places and formats and links to tax portions of same spreadsheet.
You have to watch how you structure your data rows and columns, especially when trying to query (part description to get a part number). You have to order the columns as if a person was reading it. If you are looking for a part number, when using just the description, the description needs to be in the first column, a category column would be even better category>description>part number, instead of the other way around, I was getting some hallucinations the first go ‘round, with my initial format.
Thanks for this. I was trying this function of chat gpt before it becomes a separate gpt but I was not impressed by the results. The deduction from the exel file were pretty obvious. I should retest this. Thanks!
The growing capability is interesting, but I think it’s being way oversold in this video. You make a point of saying you could make a pivot table but who knows how to do that. Any competent data analyst or anyone who’s worked with Excel at their jobs for more than a year can. It can be done in under 2 minutes and give insights to 10 questions, 6 of which you didn’t think to ask until you actually started looking at the summary. And that’s the key: the insights and questions come from the process of exploring the data. Just asking the questions you’ve already thought of will only lead to the limited answers you thought you needed, not the much wider set of answers the data can tell you.
Corporate data is way too proprietary for large orgs to ever allow it to be shared on this application. They would likely forbid this to be used as a standard application on work computers as well. I can only see this benefiting freelance data analysts.
Yeees, but... LLMs have inherent weaknesses that make their reliability at least questionable! Answers from ChatGPT are sometimes funny and sometimes simply wrong. So, I'm really not sure if LLMs are the right technology, when you need results that are ALWAYS precise, true and reliable.
Yup. ChatGPT makes stuff up unless you tell it not to. However, since in this use, the data is all self-provided, I question if AI halucinations happen in this use.
@@chipcook5346 The LLM needs to understand and correctly interpret your language. It needs to understand your column names and map it correctly as input. It also needs to find the formula terms, and so on... there are many things that are not under your control and for which the LLM has to use it's giant neural networks that contain a lot of other stuff that is not related to your problem and may interfere with your intention. But it's also correct, that because of the limited domain of data analysis, it is much easier for the LLM to avoid halucinations and shortcuts. But still I wouldn't trust it if critical decision are to be made based on its results.
@@Apenschi Me, neither. I'm very serious that this will go as the law -- and medicine -- have been going. For that matter as middle management as defined prior to the 90s went. Remember when IBM (we didn't even realize that IBM was getting bashed on the rocks) let go a huge portion of the tax base of, I think, Bryn Mawr by dumping 10,000 middle managers? At least nowadays, you won't necessarily have your entire life destroyed, and you know that your job is always on the block. At least for now, there had better be a skilled analyst involved.
Is the "Data Analyst" tool available within the current GPT-4? I do have a premium subscription and access to the custom GPT's. So my question is: If I were to upload an Excel file to the GPT-4 chat, would the results be the same as the results inside Data Analyst?
Last night I was reviewing companies I could contact to analyze data I receive from small market surveys - now I have an alternative that is less expensive per month - good for me ----however I ponder AI impact on various industries as there are currently many small to large companies EMPLOYING many people who provide data analyzing services. We as an economy are in a dramatic state of change ----- an opportunity? Thanks for a well done video.
From what I’ve heard it can only do this for a couple hundred lines of data at most. That’s hardly a large dataset, it needs to be able to analyse 1M lines of data.
2:24 I was wishing you would click on the Clip to see if it any other option such as a Google Sheets instead of just Excel or what other format type there were, or maybe to pull a file from Dropbox, etc.
Does this mean that sql is dead if we can query a database with a simple, more English friendly chat prompt? I would assume that gpt can even edit data in rows and columns too?
@@BigGuy10Points LOL "has been" not anymore. Last 1-2 years everything changed and will change faster. RIP SQL and Excel and Data Analysts and video editors and graphic designers etc.
ChatGPT is wrong many times, even on simple calculations. It's not a reliable tool. It just says "sorry" and tries to calculate again if you confront it...but if you don't know he is wrong, well...all your data will be wrong. Not a reliable tool.
I know its not a great thought, but these systems will only get better, writting their own reward functions, autonomous agents (by end of year this will change the face of ai), reasoning skills will improve and are. Error rates are dropping fast. Question 1. Why would a company pay a human to do a task that takes them 1.5hr at human wages, when an ai agent can do it in 30 seconds and cost 14 cents in tokens (server compute)? My daughter will be finishing school in roughly 20 years. Another 20 years of development - the writing is on the wall. These tools won't just improve productivity - they will replace human productivity. This is where crazy ideas like UBI come from (Universal Basic Income). Expect more protests like the Hollywood writers strike, as more and more industries are affected and the systems improve on two fronts 1. Reasoning. 2. Fewer errors.
Go from Excel novice to data analysis ninja in just 2 hours: kevinstratvert.thinkific.com/
Get the latest high-quality tutorial and tips and tricks videos emailed to your inbox each week: kevinstratvert.com/newsletter/
Please don't sensationalize titles. It's essentially clickbait.
Problem is data privacy. We have been instructed not to upload any company data on ai sites. Lets see how they solve them.
Us too
That is the business model of OpenAI and other upcoming AI companies. They will offer businesses customized, closed and secure AI engines that could only be used within the confines of the company's network.
Still dont recommend.
Exactly! This is useless because of lack of data privacy.
gpt team is private, just upgrade it
Excel VBA is still the most widely use tool for engineering and finance. Wait until you have a question from an external auditor. You still need to know the underlying process. Engineer questions are not that simple to infer. For example I just asked co pilot why the coefficient of correlation doesn’t display correctly in charts sometimes and he didn’t know why.
appreciate you shared the idea which most of us had never looked into :)
A cholera !! W ogloszeniach o prace ciagle chca SAP i ERP !!!
what if the auditor is an AI?
If you think excel is dead you don't know excel
Not excel, manual or formula data analysis is dead already. Excel is a data structure in tabular form to manipulate. All the power of analysis is on the AI side soon which will be faster and trivially done by anyone without knowing Excel. You can store all the data you want in it and just cause it's a table, it will survive as a data saving format. But not for much longer.
I understand both points of view, new technologies have been trying to kill excel for more than 20 years now unsuccessfuly... But maybe this time is different?
But also notice that when asked why profits when down the last year, the AI suggested different reasons without providing the correct answer...
This will mean that the data analyst will still need to have skills in excel, powerbi or other data analyst tools to drill in and find the currect reasons...
I believe that the data analysts that will survive the AI apocalypse are the one that can automate and speed up their work working WITH this new AI tools, no against them...
We have done it before. When powerbi appeared, people that adapted and learned and added that tool to our arsenal did very well...
AI could be another tool that data analyst power users can use to become faster and more skilled in our job...
Beside, you still need to ask the right questions to the AI and decide on the best provided solutions, and that is exactly the kind of things that good data analysts can do very well...
Excel has huge problems. You cannot use it in git. You cannot test it in a god way. The amount of data is limited and its slow.
The reason why its still in use is, because people with an economics background need bullshit jobs.
Thats why ai systems won't kill Excel
Is it dead lol
If you think this UA-cam video title is an accurate representation of what the creator thinks or what the video is about and not just a clickbait title then you don’t know UA-cam
O yes, the demo with this perfectly cleanly structured 100 row data. Gaining insights is 20% of a DAs role.
@@dreamcatcher3748 Getting the data sufficiently cleaned up and formatted to be able to glean those insights.
Structuring the data is the other 80% 😅
"Data Analyst" as a role is misleading. You're a Data-preprocessor with a side of analysis.
Ah, the old clickbait title. Even this guy does it. AI influencers do it a lot.
The tool is pretty good at cleaning data. I have been helping a company clean up their database with 30 years of junk!
This is fantastic for ease of use, but slightly terrifying as an analyst watching a tool created that can allow entry-level hires to complete high-level tasks with no actual knowledge (which were prerequisites for hire initially)
I suppose my biggest solace lies in the fact that I’m employed in a secure facility. The data we work with requires security clearance to access it, so I don’t need to stress about losing my job to some yahoo who used this tool to “look good without learning” since we would never be permitted to have Chat-GPT on our devices or export a file to use on a personal device.
@@PolyInProgit could change drastically there are LLMs which are being developed which will run on premise servers , training will be done on Large Datasets but no data will be sent outside 😂 no one is safe in AI revolution
Good, triumph in darkness, assuming there's no Sun at all 😊
Sorry, but what kind of „high-level task“ were presented here?
It can allow entry-level people to make huge and horrifying mistakes because they can't tell what they are looking at, just like highly automated industrial equipment can be run by a noob, but...
Chat GPT is a wonderful tool for somethings, but I for one would not be prepared to upload my very sensitive financial data to anywhere in the "cloud" to anyone for any reason!
Or simply mask out the PII data before you upload. Hell, I can upload a financial spreadsheet with all kinds of made up numbers.
The cool thing here is that it makes it easy to learn the tools faster and get personal projects dine in time.
As someone who is an analyst, this is both terrifying and amazing. However, your data is very clean. If you give it some data that’s dirty, it does a terrible job and will fail.
How are you so sure that's what will happen?
Because I’ve tested it. I’ve tested its ability to clean data and it usually ends up screwing it up. It’s not optimized for cleaning data right now.
@@Dragontbone oh ok
The thing is, dirty data is data full of bad patterns like null rows and columns. If you ask me data cleaning will eventually be a breeze for AI.
Today is the worst AI will ever be….
...this won't replace data analysts because you still need someone to curate and collect the data and make certain there's no discrepancies in the real world. If anything, it makes a Dat analysts life easier. You're always going to need a human to communicate. Experience is key.
AI will communicate it better than a human. I think you are overestimating humans in that respect. It's early days for AI and it moves and improves faster than we can imagine.
you don`t need to paid full price anymore
It might not replace human analysts per se but it will definitely have an impact on how many analysts it takes to do the same amount of work.
If AI tools help reduce the need for technical know-how and allow analysts to do a piece of work 2x-3x as fast, that increases the pool the people who could do analysis AND reduces the number of analysts required, a double whammy when it comes to the job market for analysts.
@@thisismarkbro no... I'm not a data analyst... But my prefomance is tied to my pay... If i get paid less than I'm not going to give my best
@@distilledgogilbaThere is hell lot of big data to be analysed .. Excel in this era is totally outdated as it hardly handles a million row data .. You need skills to work on cloud, build ETL to automate process and connect to a BI tool which again comes with set of measures to be created, BI has to be optimised to perform stutter free especially on realtime big data.. plus data handling comes with set of security that needs to be handled.. First understand what data analyst do and then think how AI will help in certain tasks.. I use chat gpt to correct my DAX measure or M codes, create customised plots through python, understand business problems if it's out of my domain and many more.. The one showed in video is very basic for industry standards.. Don't get panicked or panic others who are passionate in this area..
Chat is really useful. It’s even more useful if you have some sort of background in what you’re trying to excel at (no pun intended). That’s why people are so scared to use it, it can be used for almost anything but most people don’t have the necessary information and knowledge to maximize its use.
This is true, but these functions are better exploited by analysts because they have the knowledge of how to perform analysis and what type of analysis is most appropriate for various cases.
Instead of a data analyst you now need a fancier-sounding and more expensive prompt writer. Progress!
The value now comes from knowing how to ask the right question no longer how to use the tool.
Domain knowledge+Prompt engineering. Thank you Kevin 😊
ChatGPT told me to take a break and come back later. Guess we all need a breather now and then. Amazing video btw, thanks!
Most of the time that happens I’ve hit a dead end. If you start over and summarize what you gained from the last conversation it gets going pretty well
Can confirm. I threw a sheet with different daily rates for 3 roles in 3 tiers each by staffing companies ( procurement data) . WIth the initial prompt "Which company is the cheapest" and a couple of replies to GPT asking me I arrived at a list cheapest to most expensive on average. WIld.
Great video! Companies paid their price to avoid hacking and yet many are going to pay to expose themselves with AI.
this is simply insane ! makes works so smarter and efficient and helps in free up time to pick up some more pressing tasks . thanks, Kevin, for sharing tips & tricks along with tutorials. You have touched many lives .
I completely agree. I believe the future of data analysis will not primarily hinge on proficiency with tools like Excel, SQL, Python, or R. Instead, it will revolve around effectively communicating with AI systems, such as ChatGPT, and other emerging technologies. Mastering AI language will become a crucial skill across all professions. However, a solid understanding of your specific professional functions will be essential. It will enable you to craft precise prompts, ask relevant questions, and identify potential errors in AI-generated responses.
After seeing this video, I can confidently say excel is alive and well. I work in excel for a living and I am probably one of its most proficient users.
A user can own Excel and can use it offline
Nor are they required to upload/share their data to some 3rd party, chatGPT.
Still very useful,, especially when working with data that isint sensitive
Great video. Next, show how to use it to generate files (PDF, PowerPoint, Excel w/Charts/Dashboard, Power BI) to convey the information!
Great video!
You missed to show that you can ask GPT for a downloadable png of the graphs
You can also do so much more, like flowcharts, mind maps, heat maps, etc
Good point!
@@KevinStratvert In my case, I had a sheet with several thousand rows of Amazon sales data, including daily and hourly data. GPT was able to analyze it and give me deep insights, such as hourly and daily trends. Plus, it provided me with super cool graphs to show YoY comparisons and hourly sales heat maps too. I was super impressed and actually found it very useful.
I always love the "If you'd like to follow along" feature to videos and you've added this and done well with it! Thank you for sharing! Let's connect
Thinking on this more, beyond the magic of seeing the GPT engine produce analysis, the Achilles heel of this tool is its single use and lack of memory. There is no memory after the session of questions. Every spreadsheet analysis is one off. Super the boss might think it’s amazing the first time, but as soon as the spreadsheet changes - and it always does - you’ll have to start the chat and all your questions and charts all over again. Suddenly it doesn’t seem such a time-saver. Meanwhile, if you get your data person or the Excel whiz 2 seats over to make a pivot table, chart, or a full data model and automated dashboard, they’ll update instantly as new data gets put in, changed, or corrected. Waaaay better for anything more than the most simple of questions any business not started literally yesterday should already have answers to.
It’s great for simple data, would be great self serve for basic request…feed it specific data
I remember my excitement in 1999 when Google Search became open for the public at large. Haven't had that same feeling in 25 years until now with ChatGPT 😊
Going from Altavista to Google was not that huge of a leap like GPT is lol
@@betimz I never used AltaVista.
I think it's nice that AI is able to read through spreadsheets and derive some meaning, but it's still a far cry from actually working with the data. These tools are still evolving not able to handle complex or dirty data and have limits. It's the same as using AI for any other work, you must be aware of what it knows, ask your questions carefully and double check the results. At the moment the only danger to anyone's job will come from managers who don't understand the job. It's up to you (the data analyst) to make them aware of what you do over what the tool does. Turn it on its head and show how the tool could help you but not replace.
Yeah once it starts interpreting dirty data and cleaning it up like a good human analyst does but in 10 seconds vrs an hour for a human that will be cool to see
To be honest it looks good, but i would save time if i use directly a pivot table and insert a chart, at some point you need to do stuff by yourself and not always rely on AI.. :)
Exactly. This IS done better with software. And you can save that file too. You can’t save the output from Chat GPT because it doesn’t produce a saveable file.
Indeed. This can help some total novice, but nothing showed here looks like a time saver, or perhaps those verbal interpretations could be if I write bullet points/hypothesis and ask gpt to elaborate from those.
You could also upload screen shots of analysed data whether its tables, charts, or anything else and ask it to analyse it for you. It's a revolutionary learning and productivity machine and I love it 😎💪
That said, it might make mistakes sometimes. So, you should always be vigilant, careful, and even double check the information if you suspect something isn't accurate.
It doesn't just do Excel but SPSS, JASP, R, Python and much more. Go have fun with it and happy coding 😁
it's sad and here I am learning excel hard this past week
Same here.
Do not worry, it is ok to use AI for easy use, but when you have data with thousands of row and columns, even AI cannot accurately analyze that. Imagine even excel crash with those types of data and that is why you would need Power BI or Tableau. Second and maybe it should have been first is privacy of data. There are a lot of companies that don't want their data put in AI. And they clearly specify to not use AI.
Dude just gave all his company's secure information to bigdata and now all his competitors can have access to it for a price. Later, the Cookie Company went belly up.
Things will go to next level when it cab do things for you in excel like making dashboards, etc
If you upload an Excel file into ChatGPT as highlighted in this video, would you not worry that all the data within the file becomes part of the AI, and therefore part of the whole AI realm? If that's the case, then your data is part of the AI realm, and might be accessible to/by other users.
Where do you save your Excel files?
Can ChatGPT do cluster analyses or similar? That'd be really helpful. If it can, do the same data assumptions apply, or would it be able to recognize clusters without making sure these assumptions are met. Thanks, Kevin. You rock.
Damn! Mind blowing
About to show my boss some levels
Chat GTP sounds good in theory, but when you need to load your own data into it, (using a spreadsheet) as how your supposed to avoid using (because you said it was too cumbersome, no one wants to use formulas) Where doss your info go, how it's used, who has access to it. Where is the privacy?
Super vid! I uploaded your KCC to gpt and asked it a bunch of questions. Very helpful results and useful insight into how far this can go.
It can also conflict with customer privacy and personal information laws. Once it is on Chat GPT personal and private information will be compromised😢
So amazing. People/businesses will need to encrypt their spreadsheets. Are we ready for this?
That last quote was a banger Kevin!!!
“You don’t need to know how to use the tools , you just need to ask the right questions and to make the right decisions.”
3 million subs generating just 400 comments on a video published 3 weeks ago. You are good at IT.
In the case of a Data Analyst's employer - company data, that is sensative. Employers expect you to keep their data within thier intranet or approved vendor portals. You will not be able to drop any such files into this tool, sharing it with non company systems is a violation of InfoSec policy.
Wow! This video blew my mind! ChatGPT's Data Analyst is a game-changer for data analysis. Thank You Kavin!!
If you have experience in data analysis, you will know this is pretty basic stuff and it’s far from being a game changer
wow truly instructive videos do exist after all, wow. props bro.
Yes, you are right, its truly incredible.
It’s great. I wonder though, about privacy and data governance issues. Here in Europe, we have pretty stringent rules on data protection like GDPR
Very cool! I enjoyed it. Art of asking is going to be valuable.
🎯 Key Takeaways for quick navigation:
00:00 *🤖 Introduction to ChatGPT Data Analyst*
- Introduction to ChatGPT's functionality as a personal AI Data Analyst.
- Overview of premium subscription features and custom GPTs.
- Demonstrations of various GPT capabilities, including data analysis and visualization.
02:16 *📊 Data Analysis with ChatGPT*
- ChatGPT's ability to analyze data from Excel files.
- Demonstrations of simple and complex data analysis queries.
- Explanation of how ChatGPT aggregates data from multiple sheets and performs calculations.
06:19 *💵 Data Visualization with ChatGPT*
- ChatGPT's capability to create visualizations, such as bar charts, from data.
- Demonstrations of modifying chart properties, such as axis formatting and bar colors.
- Ease of making visual adjustments through simple text-based commands.
07:42 *📈 Advanced Data Analysis Queries*
- ChatGPT's ability to provide detailed analysis and insights for complex queries.
- Examples of querying trends, identifying potential reasons for flat profits, and analyzing seasonal patterns.
- Illustration of ChatGPT's potential to assist in decision-making processes and business insights.
Made with HARPA AI
Sir, Please give a video for learning Business Analysis with Excel.
If you don’t want to pay, just use Python and R… the codes for data analytics are done, you don’t have to think for the input. Only for the output.
If there is something complex, it is the real life job, not how to code the analysis.
Will it coordinate an in depth analysis of internal and world event driven items such as the loss of a key employee, bringing on a new vendor, a flood in the region, the Suez Canal impacting just in time delivery dates, changes in laws or government, multiple data souces… and things you have not considered, when making predictions?
I gave ChatGPT 850 lines of data and when the results didn’t make sense she apologized and said she only used a sample of the data and then gave me the correct results using all data. The fact that I had to do that made me not trust ai right now.
Can it do data cleaning? Or can it analyze raw data, data thats not yet cleaned?
You’ll just have to explain to it what is “cleaning” and make some samples for it to know what do you mean and then prompt it to work on it.
Yes, it can clean the data, but it's not perfect. You'll likely need to guide it, but it can pick up on the basics.
as a former Microsoft employee! i now feel good I quit my job as well! I used PowerBI when PowerBI was a God in the industry! now what can PowerBI be with this AI capabilities
I’ve realised that people who don’t like AI starts their process like this.
Stage 1: AI won’t be able to do XXX.
Stage 2: AI won’t be able to do better than humans in XXX.
Stage 3: WE SHOULD REGULATE THIS ASAP!!
This is a really interesting video. Nicely done. I echo another comment about data privacy, but this seems worth exploring more.
Sumif equation for the revenue for one customer, reports though with tables and graphs definitely could hasten doing analytics on a per workbook basis.
Excel is far from being just an analytics tool. I've been using spreadsheets since the days of the old paper columnar pads, then Lotus 123, and then Excel. Excel and the others are often points of original entry for data. So, no, Excel isn't dead; it will be around for a while longer.
It’s not that important how chat GPT figures out how to make a poem or picture. But if you are relying on chat GPT to look through a bunch of spreadsheets and say - “your company is profitable” or “it’s losing money” I wouldn’t just say “Ok! I’ll go with that and make a presentation to the shareholders now”.unlike other things in AI you need to know how chat GPT is coming up with these answers.
But can it add data to the spreadsheet? I'd love to have it as an assistant that you could tell to add a new stock purchase to my existing portfolio in all the right places and formats and links to tax portions of same spreadsheet.
You have to watch how you structure your data rows and columns, especially when trying to query (part description to get a part number). You have to order the columns as if a person was reading it. If you are looking for a part number, when using just the description, the description needs to be in the first column, a category column would be even better category>description>part number, instead of the other way around, I was getting some hallucinations the first go ‘round, with my initial format.
For someone like me who has to analyze a pile of data in a short time, this is gold.
curious how it has worked out for you. I have 12 tab sheet that I want to run breakeven analysis by unit, by location, by headcount, etc
Thanks for this. I was trying this function of chat gpt before it becomes a separate gpt but I was not impressed by the results. The deduction from the exel file were pretty obvious. I should retest this. Thanks!
Very cool and useful for my data science career
Now anyone can be an excellent analyst.. even someone who is not an analyst..
I dunno, as an Aerospace Data Analyst, ChatGPT replaced some logic in my DAX logic when asked to make it look pretty and it’s terrified me ever since.
Thank you Kevin I hope your company makes even more profit.
'ask the right questions and then make the right decisions'
The growing capability is interesting, but I think it’s being way oversold in this video. You make a point of saying you could make a pivot table but who knows how to do that. Any competent data analyst or anyone who’s worked with Excel at their jobs for more than a year can. It can be done in under 2 minutes and give insights to 10 questions, 6 of which you didn’t think to ask until you actually started looking at the summary. And that’s the key: the insights and questions come from the process of exploring the data. Just asking the questions you’ve already thought of will only lead to the limited answers you thought you needed, not the much wider set of answers the data can tell you.
Most companies data could not be uploaded to unsecured platform so we will have to keep leaning the hard way
Hahaha don't worry they will find secure ways to do this. It's improving at a crazy pace right now.
Awesome demonstration, thanks
Thank you Kevin. You are just amazing.
EXCELENTE VIDEO HERMANO, GRACIAS POR LA INFORMACION.
Great... Im just learning to be a data analyst and I see this. FML
nope, you are not 😉
@@vlastimilpetr9630 he is not what?
Find something else. No need data analysts anymore.
By watching your videos I can understand why Data Analyst jobs will be gone soon 😢
New sectors will emerge and existing people will have to suffer.
I have found extraktAI to be a lot better at this and so much more user friendly. But you explained this really well tho, thanks!
When chat got or copilot gets into corporate world built in it is game over for data analyst
Hardly. Can't see AI ever building pipelines and producing a clean data set from a mess coming from an entirely different system.
Corporate data is way too proprietary for large orgs to ever allow it to be shared on this application. They would likely forbid this to be used as a standard application on work computers as well. I can only see this benefiting freelance data analysts.
and small companies who dont have the luxury of being proprietary
But, to ask correct questions you need to know Excel. So, don't give up on learning it
Your channel is one of my few, very few favorite subscriptions on UA-cam ever.😺Straightforward. With substance.🎀🏆🏆
Yeees, but... LLMs have inherent weaknesses that make their reliability at least questionable! Answers from ChatGPT are sometimes funny and sometimes simply wrong. So, I'm really not sure if LLMs are the right technology, when you need results that are ALWAYS precise, true and reliable.
Yup. ChatGPT makes stuff up unless you tell it not to. However, since in this use, the data is all self-provided, I question if AI halucinations happen in this use.
@@chipcook5346 The LLM needs to understand and correctly interpret your language. It needs to understand your column names and map it correctly as input. It also needs to find the formula terms, and so on... there are many things that are not under your control and for which the LLM has to use it's giant neural networks that contain a lot of other stuff that is not related to your problem and may interfere with your intention.
But it's also correct, that because of the limited domain of data analysis, it is much easier for the LLM to avoid halucinations and shortcuts. But still I wouldn't trust it if critical decision are to be made based on its results.
@@Apenschi Me, neither. I'm very serious that this will go as the law -- and medicine -- have been going.
For that matter as middle management as defined prior to the 90s went. Remember when IBM (we didn't even realize that IBM was getting bashed on the rocks) let go a huge portion of the tax base of, I think, Bryn Mawr by dumping 10,000 middle managers? At least nowadays, you won't necessarily have your entire life destroyed, and you know that your job is always on the block.
At least for now, there had better be a skilled analyst involved.
Is the "Data Analyst" tool available within the current GPT-4? I do have a premium subscription and access to the custom GPT's. So my question is: If I were to upload an Excel file to the GPT-4 chat, would the results be the same as the results inside Data Analyst?
I’m interested in the answer to your question
Is Google AI able to do this in Google Sheets as well ?????? Great Video !!!!!!!!!!!!!!!
can ai help understand general ledgers in excel for accounting?
Yes, just tried it, it's insane, my company is sending us to Texas to get some AI training lmao
GPT is IFFY at BEST for data analysis. I’d give anything of it worked better.
Hi Kevin, can you do another video of Excel with Copilot Pro or Copilot for Microsoft 365? Want to see how it compares to this
Yes, truly incredible thank you, Kevin
Last night I was reviewing companies I could contact to analyze data I receive from small market surveys - now I have an alternative that is less expensive per month - good for me ----however I ponder AI impact on various industries as there are currently many small to large companies EMPLOYING many people who provide data analyzing services. We as an economy are in a dramatic state of change ----- an opportunity? Thanks for a well done video.
From what I’ve heard it can only do this for a couple hundred lines of data at most. That’s hardly a large dataset, it needs to be able to analyse 1M lines of data.
if you could do more of these, super helpful
2:24 I was wishing you would click on the Clip to see if it any other option such as a Google Sheets instead of just Excel or what other format type there were, or maybe to pull a file from Dropbox, etc.
That is now available! You can upload from Google Sheets and OneDrive!
Does this mean that sql is dead if we can query a database with a simple, more English friendly chat prompt? I would assume that gpt can even edit data in rows and columns too?
No sql has been available for a long time.
@@BigGuy10Points LOL "has been" not anymore. Last 1-2 years everything changed and will change faster. RIP SQL and Excel and Data Analysts and video editors and graphic designers etc.
@@speedymr it just makes their work faster
Awesome video! Thanks so much!!
oh wow Kevin Doing great videos again . Thanks for your hardworking for us!!! Super Video I needed
Brilliant and it is very useful Presentation.👌
ChatGPT is wrong many times, even on simple calculations. It's not a reliable tool. It just says "sorry" and tries to calculate again if you confront it...but if you don't know he is wrong, well...all your data will be wrong. Not a reliable tool.
Kevin was actually generated by Sora, he's been Microsoft's Turing test all along.
I know its not a great thought, but these systems will only get better, writting their own reward functions, autonomous agents (by end of year this will change the face of ai), reasoning skills will improve and are. Error rates are dropping fast.
Question 1. Why would a company pay a human to do a task that takes them 1.5hr at human wages, when an ai agent can do it in 30 seconds and cost 14 cents in tokens (server compute)?
My daughter will be finishing school in roughly 20 years. Another 20 years of development - the writing is on the wall. These tools won't just improve productivity - they will replace human productivity.
This is where crazy ideas like UBI come from (Universal Basic Income).
Expect more protests like the Hollywood writers strike, as more and more industries are affected and the systems improve on two fronts 1. Reasoning. 2. Fewer errors.
I personally find this way more groundbreaking than Sora.