7 Ways to Use ChatGPT for Analytics | WebinarApr 25, 2023
The three GPT buzzword chat is a word on everyone's lips and we are so grateful to welcome Alex Breberg back to the channel. welcome everyone while I'm here I'm William the event and communications leader here at the Foundry race to everyone joining in let's just give a couple of minutes for people to join in but at Big Market you can see we've got the chat activated. the right side maybe because it will all be about the GPT chat just say why you are interested in the GPT chat where you are joining from and what is your name also if you are viewing on LinkedIn also if you are viewing on YouTube I know Alex al
waysbring your people, we love to see them.
We've done a lot of workshops in the past and presentations so I'm thankful to everyone for joining tonight and it's about seven
waysto use a GPT chat for data analysis as you can see in the big market we have some emojis too press those, let's know you're here we want to make this as interactive as possible before we get started let me introduce Foundry career very very briefly um Foundry career is the online school for your career move into technology and We guide you from a complete novice to a job-ready professional in data
analyticsand help you land your job in the field.
We are not old school. Our programs are so flexible that you don't need to quit your day job, you get regular mentoring from not one but two industry professionals, so you're a mental coach as well as a tutor. That's our dual tutoring model, and if you don't get a job within 180 days of graduation, we'll refund your tuition in full. that's the pro casting guarantee of work but i don't want to talk too much about pro casting if you're watching on YouTube and you're interested in doing a pro cap show we've added a book, cool link at the bottom. to one of our program advisors so if you're interested in the right term if you're interested in the Foundry career book with a program advisor and at the end of this
webinarwe'll have a live meeting so please leave your questions big score LinkedIn YouTube you have a lot of team in the background and we're going to have a great session q at the end Alex I really don't want to say anything else this is all about you it's all about gbt um I'll give you the floor awesome thank you so much and yeah thanks to the Career Foundry for putting things together.
It's always so much fun. es, we have a very important topic today and of course I'm going to talk about how we can use gbt chat and data analysis, but I'll also talk a little bit about other topics related to gbt chat and I'm sure in the end during q I already know some of the questions that are going to come up um and i'll stay a little bit longer a bit longer uh if we have a lot of questions on that and then um let's move on uh let's kindly dive into this i'm going to start by saying that i um i definitely fall a little bit in love gbt chat if i ramble too much i apologize um but i'll try my best to keep it short uh well basically how are we going to do this.
I'm going to have some slides that explain some of the ways that we can use it. I'll talk about this and then I'll show you a very quick demo. Chachi BT of how you can do these things um I won't do it again I can't do a full Deep Dive demo um but I'll just show you a few things and I have a whole video on how you can do that if you want to watch it on my channel too but this It's me, it's Alex, and let's talk about gbt chat, so if you don't know anything about Chachi BT, you're just hearing it for the first time. time uh it's pretty awesome uh it's basically uh a chat where you can talk to this computer and it will give you answers in a very simple sense but to be a little more specific it's built off of a great language model huh these Movies are relatively new uh only going back four or five years but recently started to grow because of Chachi BT in November last year uh really intended to help understand a lot of things just talking to them naturally it's been integrated into a lot of different things even chat gbts released something called AI whisper where you can actually talk and communicate just using your voice which is pretty amazing um but most people do are using right now just typing in the chat and doing it that way um and it has a ton of different uses now these three uses are like the tip of the iceberg there are literally thousands but these are three that i think everyone might be fairly familiar with ease uh you know look um you can use it for language translation almost like a google translate you can do that uh in Chachi with tea you can use it for things like content creation that's something I use it for um you know I think you know here there a topic and then i asked chat gbd hey what other topics could you talk about on this?
It just gives me ideas so it helps me create content and then data analysis which of course we're going to focus on today. Let's talk about how it works. I'm going to talk mostly on the user side, though if you really want to like it, go crazy, you know, look at how it works in the back end of um on how these big language models work because they're crazy, they're super cool, basically how it works is you have this chat window and you ask him to do something now you can ask him a question as a fact you say do you know in 1997 how many people lived in the United States and you can probably get that information back or you can do something a little bit more advanced if you gave it some context you say hey you are a data analyst you are a python expert i want you to write this code or something so you can you are asking it to do something you are giving it some context giving it some information now Chachi BT It's going to take that information, it's going to process it, it's going to do some kind of back-end reasoning, it's going to generate an output and what's really cool and what's become such a big part of EBT chat is that he has some kind of memory, a working memory of previous conversations, so it's not just asking him a question, you're having a conversation with him and you're saying hey three questions ago I gave you this information I have a follow up question on that and it tracks all of those things and it's pretty awesome so we're going to talk about these seven ways and again I want to preface this by saying these seven ways are very specific to data analysis the more technical side of things I'll mention at the final a few more ways i've been using it and i've been using it a lot so like you know these are good ways to really start using it things you can definitely start using but then i'll give you other ways in the ones you can use what uses it too, the first one is just learning yourself when you know you're First you're starting out and you're trying to learn.
You are taking courses that I highly recommend. Sometimes you get stuck and you're just like me. um, you can ask Chachi BT to walk you through this, show me how this code works and he'll explain it to you, um and you can even ask him to create some kind of lesson plan like he's trying to learn. SQL, what should I start with? I don't want to ask Alex, he doesn't know anything. I want to ask chat gbt because he knows everything so just ask him for a custom plan. You say I'm a new data beginner.
analyticsi want to learn SQL how can i do that and it will build it for you and you can ask it for links and the place is where you can study certain things and again it will offer those links and resources quite amazing um I'm sure a lot of people are getting into analytics or taking courses or doing all of these things so this is just another resource to start learning um or even if you're looking for more advanced stuff even uh can be really great at teaching yourself next thing is explaining the code huh this it's pretty it's really cool honestly huh because even for me even though I feel like I'm pretty advanced on a lot of you know how to write SQL and Python and even some .
I want to use python or I want to put this somewhere else. Let me change it to python, so I'll say you know how to convert this MySQL. code in uh you already know python code and it does a very good job or at least gets you 80 to 90 of the way and just saves you a ton of time which is pretty good next thing is answering domain questions now i worked in healthcare for many years huh I love healthcare I think it's fantastic and for data analytics it's actually a little bit underutilized um huh data analytics is underutilized in healthcare and what has been really cool it's just this understanding that Chachi BT doesn't know everything but they probably know more than you do so for example I worked with claims data and if you don't know healthcare in the US or you are in another country the The health care system is not great, but what I will say is you.
I know I worked with claims data for a long time and a lot of people would come to me and ask hey what do I need to know about claims data for data analytics because there is this data analyst job that is open for claims data. and i would say a bunch of different things like i could talk all day about health care and claims data i just know too much about it and now you know instead of doing that i'll say you know let me real quick let me ask to gbt chat. We'll get in touch with us and I'll say here's the real stuff and here's some resources and they'll provide me with all of that stuff um and I'll keep talking to them and I'll still like to help them but at the same time you know you can give you a lot of context on how different domains use your data Healthcare is just an example but let's take Finance you can tell how you know stock price analysis using python or MySQL using this data type or that data type you can really use it as just a expert in the field and again not perfect sometimes he does things a bit wrong but for me uh forplay around with it and test it especially with healthcare which i have a lot of knowledge in it's pretty good it's pretty accurate the following is commenting your code now if you're just starting no one is commenting your code like no one no one is commenting the code huh that's because you're just starting out but when you get into a real workplace I promise people will be mad at you if you don't come with your code at least they did on my team because we go through a lot of code you know I would write my procedure stored in MySQL and they get pretty advanced and then I'd like to pass it off to a database developer and he's like what the hell did you do he's like this is this spaghetti code here and I'm like yeah ok I did good like this that you do your job and then you know it would fix it but almost every time it was a requirement as part of the team to comment your code the dbt chat does a very good job of this where you can put your block of code and you can say add comments to this and it's almost like when I was talking earlier about explaining the code it's like instead of explaining the code to you I now add comments to explain the code in your actual code so you can copy and paste it again really very useful and what I will say.
Know? Commenting out the code is kind of put down, like no one really takes it seriously. you, um, so there should be excuses in the future, we should all comment our code. I'm trying to push that into the data analyst community because I think more needs to be done, um, okay, let's move on to the next one. one is data cleaning now data cleaning is when you take raw data from a source and clean it up to make it more usable for the final product or final analysis so if you haven't done that before that's ok really It's just taking messy data making it more usable and I've really spent weeks of my life cleaning up one or two sets of customer data because it gets really complex, um, that doesn't just come from the data being messy right?
It doesn't mean that the data is, um, all misspelled and misspelled, sometimes it's more business rules, um, where the client says yes, do that, except when these and these business rules are what Chachi BD normally does really can't do. it won't be able to understand those business rules the way you do because you're working with the client, you understand how they get their data into the feed, but what it's really good at is making data cleansing more obvious. that you're going to have to do anyway, like misspellings or formatting issues, those are things that you know just take a lot of time to have to search and find and use, uh, they're fixed, so now you know that you really you can do it. in a small sample of data if you are using the EBT chat window at openai.com right now i am going to go off on a very small tangent stay with me here i apologize now once i get going i can't stop right now microsoft is integrating a lot of this gbt chat into its systems.
I'm sure you've seen it with Bing with their search engine, but they're also going to do a lot with Azure if you're using anything with Azure likes their Azure data factory, Azure data lake, Azure data store, Azure data, all of those things are going to have these kinds of things built in in the very near future, if not, some of those products can already do it and therefore on a small scale. now everything but you'll be able to do it on a large scale in the future um probably not too long where you can say hey I've got these millions of rows not just 10 rows you can do in GPT chat but I've got these millions or 10 million or a billion if you're working with big data and you can say hey you know what are some things you should identify to clean up here and then you can ask it to clean up that data um there are a bunch of other factors that go into data cleansing like keeping the raw data using staging databases a lot of different things but that's a whole different
webinaron that process so let's keep going let's keep going I'm trying to keep it focused I've got a lot going on I've been talking for almost 20 now minutes I have a lot to get through so now we have some debugging code you know trainers don't work all the time I can tell if you've ever seen any of my tutorials online.
I'll make these long tutorials and throughout them I'll make mistakes and have to figure it out. Chaturbate you can do a lot of that for yourself and actually it's been very useful for me I've been doing um some much more advanced coding in webdev um as well as you know creating much more complex projects in web scraping and stuff so using selenium and some other libraries and it's been really complex even for me I'm like ok I'm pretty good at this but pagination and there are some issues on certain types of websites and it's hard it can really get complicated and I found that when i put this code really advanced or advance for me let me say its Advanced for me when i put it there.
I'm like I'm wrong with this. I don't understand why I don't get the result or it doesn't work. i want i'll just ask hey i debugged this code i was expecting this but i get an error and here is the error code when i asked for this info often as you know maybe 70 of the time you could actually help me fix that problem very quickly, um, that when you work alone on these projects, as I usually do, you know that I don't have someone next to me that I can meet in the hallway or on my team who I can ask to help me with this, since You know, the chat DVD has been like a little personal coding helper for me, especially with the more advanced stuff and I've been really impressed with how it has solved a problem. a lot of my problems that i had um and what's really interesting also is if i'm generating the code i can take that code and try to implement it in my current code block uh i have and if i have a problem i pull it back and return it and then it debugs it and I get it back so it's me I really feel like I'm having a conversation with gbt chat generating my code explaining my code debugging my code it's really helpful it's really fascinating um and you know if you haven't used it yet if you're just starting programming again you know it might help you initially start learning these things but when you really start to get involved in these bigger projects you know this is a great resource um I I have been using the free version a lot for a long time and you can get that you can do a lot even with the free version of Chachi PT um which is really cool let's move on to the next one I'll take. another sip of water, so here are some considerations and that's the seven things again.
I'm doing well on time. I also have a demo coming up so here are some considerations that it's definitely not perfect and I think that's a good thing huh because if it was perfect it would be a little weird but it's not perfect and it makes a lot of mistakes and there's something called hallucinations and if you haven't heard of that term yet it's definitely like it's getting very popular because uh chati will quote without quotes hallucinating facts will hallucinate um numbers even basic calculations that can go wrong and that's something Chachi or uh la Open AI is really working on getting you to do basic calculations because it's just that you have a really hard time and you can't run the code properly so you either have to have some kind of IDE or you have some limitations on what Chachi BT can actually do , but hallucinations are really important, um and I'm not. to know if they're going to go away any time soon, you really need to fact check, uh uh, fact check, what you're putting in there, for example, if you're asking for more factual information, if you're saying hey, you know, explain it.
I, uh, this is not Sequel and it gives you a result, you know I've had moments where I'm even like I don't think it's right and then I'll know I'm like because I've used it enough. I think that's not exactly what it does. I'll go look it up and it's like it's a little off but gbt chat has a way of writing it and the way the interface is where it is. it makes you look like it makes it really believable um and sometimes it's like 90 the way its accuracy knew the last ten percent is wrong and you really have to watch out for that so I want you to know that just you understand your expectations, it's not a perfect system the next thing is don't put any confidential information in gbt chat don't put your you know personal information don't put your company's personal information um Microsoft or not Microsoft um who was it was a company that had just of having a big article written about them because they're putting trade secrets and documents in there that shouldn't have been um and then you know things potentially leak and get into the system because um you know in a really simple scale is this reinforcement learning , it's learning and it's having a feedback loop and it's taking more data and it could be taking your sensitive information and using it for different information so you don't want that so uh the last thing I want you to think or remember is that gbd chat is really as good as you get if you give it a super simple message like what is this? it will give you a pretty simple answer although sometimes it can work very well but the more context you give it the more information you give it will actually give you much better answers and so the more specific you can be the more information you can feed , the better context it has as a whole and the better, uh, um, off uh, output, it'll give you, so with that being said, let's jump to my screen.
I'm going to open the GPT chat. I'll show you what it looks like if you haven't seen it before. um and i'll give you a little bit of uh just a little demo of how to use it we're not going to go over every single thing we talked about the seven points because it's going to take too long but let me pull up my screen real quick ok i hope you can see my screen right now let's see if it can be ok so i'll assume it is and if it's not it's not good let's assume it is although this is the interface for GPT chat at least they are on their website chat. openai.com now these are all my previous conversations uh don't look at this one right here i was looking for uh Darth Sidious and his background info seeing what Chachi BT knew about it um i read the forums but you know i wanted to see what gbt chat said about anyway uh right now i have PL i actually paid for the plus now i have access to three different models the default and it gives you info here some different ones now i'm going to use chat gbd4 which is their latest model I'm going to go ahead and click on that and let me hide this real quick.
What I'm going to do is give you a demo. I'm going to go over to this excel and this is some dummy data although this is accurate data um this is Tesla data go ahead and copy this and I'm not going to paste it yet but I'm going to give a little bit of context first and then I'm going Gonna paste it now I'm going to update this really real quick just to make sure it's fresh and it works so this is what we're going to do I'm going to tell you you're a data analyst you're going to analyze data for your Tesla company so this is an analyst of data at Tesla that's the context we're giving him at least um and I'm going to keep it simple because again um yes he's going to give us an answer.
I call it key, but try that gbd is going to give us an answer. You're going to understand that that's what you do, I mean, as a Tesla data analyst, my main focus is this, it's going to give you some information, now I'm going to feed it into a data set, I'm going to say um, here's a data set that you're working with and I'm going to paste this information in and you know it doesn't seem to be formatted well let's see what it says in this and then it will give you the data site we can get some ideas and do analysis on various aspects , so again it's about this Person he's a data analyst here's the data I'm given and because as a data analyst when I work with data here are some things I should be doing now often when you're working with you, you enter some information he will build this little table that he will remember for future conversations so right now he is you know or he but Chachi BT is you know go To give us some output I haven't asked him to do anything besides exactly what he's doing right now, which will do its thing now as you talk or as you type um, i'm going to talk to you about what i'm doing. to do in a moment I'm going to ask you to write some code for us and since this was in excel you also need to write some excel formulas for us and I'll keep it pretty simple so you know because if I go too far it's going to take a long time to that actually works um I could take it as you can see it's taking a while to write all this stuff so if I give you a long notice it could take a long time to generate the code so I'll go ahead and stop it here it's us giving, actually, some somewhat generalized information, but also specific information, so this is what we're going to ask, we're going to say please and I always like to be nice.
Please write my sequel. Whoops on my sequel query to see which vehicle spawned us the mostprofit after expenses and let me spell it right so i say uh i wanted to give us this information now i'm going to let you write and i want to go back up real quick so we have the price of our car the cars sold so how many cars were sold and how much these vehicles cost to produce these are really accurate this is actually um interesting inaccurate information i think though this is based on 2022 but we haven't given any information on these columns what data is there let's go back and see what it's giving us so we'll just assume that the car or the table is called Tesla sales and it's going to infer information based on the data that's in the columns as well as the names of the columns themselves so it just generated this and this gain here is exactly how it would have done super simple we have the price of the car that's how much we're selling it for less than what it costs to produce multiplied by the number of cars sold and that's our profit for that actual vehicle so either the model S or the three or whatever so if I plug this in I would just copy this code I would connect it to a SQL database if it was in SQL now let's say I had it as a panda dataframe or a polar dataframe say, um I wanted to convert this code to python pandas and I'm going to go ahead and run this and we're getting the same thing this and this happens a lot with chat gbt I'll just let you know this is when you're using it if you don't ask right away, a times it's huh no it didn't work correctly now this is a good example that it remembers this conversation because I had to refresh but it kept this conversation in memory and now I continue this conversation saying convert this code to python pandas so it will go ahead and do it which for us now this is going to take a little bit usually Imports pandas creates the dataframe and then it's going to create the code using a dataframe so give it just a second um but the last thing I'm going to do have it become an excel formula, we'll take that excel formula and put it back into our excel that we had and that's going to be the last thing that I'm going to demonstrate, um again, is that I can't express enough how I am, the board it's the word it's pretty it's magic it's just amazing um it's pretty it's pretty amazing to see it happen visually but also to think about how it works on the back end it's amazing and as a data analyst you use things like whether or not to use this you're going to spend a lot more time writing this code you're going to spend a lot more time possibly understanding your data um and so it's just that it's an enhancer of your ability and it can also allow you to do a lot more so this is the code.
If I just look at this, I'm going to take a look at it. dataframe that's taking that column that's saying minus production cost multiplied by that all this looks good and then we're even, it's even sorting it for us so, yeah, you know it's going above and beyond even in this case and creating um you know this most profitable vehicle and that's printing it. I didn't ask you to do that. Sometimes it does. I didn't ask for it, but that's what he did. Now I'm going to say. Can you convert this formula to an excel formula.
I don't know if that's the best way to put it, but let's see if you get it. Let me take a sip of water real quick. and this is a really great example of making a mistake, but, what's really cool is it's going to say you know, okay, if it's in excel, it starts in cell A1, so it's actually very interesting, now it's okay, it's giving us this really, um, I think it's converting this formula right here, which because it's using, let me see, we're solving this on the fly. Let's see, yeah, this formula isn't 100 correct, but what we're going to do is take it anyway, yeah, okay, it's converting, it's actually converting this python code into a formula, so it didn't work. exactly as I actually expected I just wanted you to give us a really simple one um but let me um there we go now oh it's uh formatted run now um I'm not going to uh I'm not going to try to get this to work because I don't want to waste anyone's time but yeah would have requested it based on mysql i think it would have worked correctly but it did it based on python code so you know it keeps everything in context keeps everything in memory i asked him to convert this formula and i'm sure he was thinking of this so those are some of the little limitations you need to be really specific in what you're asking and how you're asking it because again if you're not very specific you can often make mistakes like this um that was my fault if i'm honest it was completely my fault so with that being said you know how you can use it. this is what it looks like this is just a very quick demo but there are a lot of tutorials i have some on my channel but there are a lot of ways and you can even see their documentation on how to use it how to request all of this stuff let me get back to it very quickly .
I'm going to stop screen sharing and I'm back, so it was just a quick demo. I didn't show them everything I could have asked for. very fast um but go and try it yourself it's pretty awesome um we don't show slides anymore it's just me here we go uh it's pretty awesome um and the technology is getting better and better and better they launched us in november and it's already made leaps and bounds huh and it's already been improved by Leaps and Bounds in such a short time so here's what I'll say in a nutshell: child gbt is an amazing tool that's how I see it it's a fantastic tool and if you don't use it you're going you're going to waste a lot of time this is how i would perceive it now if you're in a job where you're an active data analyst and you're not using something like this at least for the simple stuff you can enter sensitive information you can't plug it into your database, but if you're not using it for templates that won't reveal trade secrets or personal data or information, you know I think you're using you're not using your resources effectively.
I just became a huge supporter of them. A big believer in it now, especially after using it since November. I have used it since it arrived. out um it's just an amazing awesome tool I know there's going to be a thousand questions uh um in just a second so I'm not going to go into depth um some other things I could talk about um but I'm going to pass it back to Will. I'm going to let him talk a bit then we'll ask some questions and again I'm looking forward to the question because I can talk about this all day I'm so weird thank you Alex. for that great presentation and also for the live live work on the GPT chat I think it's great to see it work but also how you only enter one part and then the second part and the third part so you get really far away from what you you started, but you see the inner workings of the GPT chapter, so I think that was very helpful.
There has been a lot of participation in Big Marker. There has also been a lot of participation on YouTube and Linkedin. If you have any question. now it's time for questions so the floor is open it's your time to ask questions about the GPT chat or I know we have a lot of people watching tonight who are just starting their journey in data analysis so if do you have any introduction, um know questions about data analysis the industry or the career Foundry this is the time to ask the questions you have a question alex i would like to say one thing i will answer any questions you have to the best of my ability i promise you but I'll say um I'm not an expert on gbt chat I actually use it a lot I would consider myself an expert user but I don't know all about it but I've used it for a thousand different things I find it awesome but you just want to put that context in there .
I don't claim to be the only source of Truth on this, but I will do my best to answer any questions you have, any questions we can. let's not answer we'll just refer everyone to the GPT chat um to ask those questions I think actually the first question I think is a great one to start with because I know Alex showed us the different versions of a GPT chat since it's been released um Bridget had a good question, do you think paying for plus is worth it as a beginner now? I don't recommend it and in fact I didn't pay for it until about two months maybe it was a month ago it was recent and I only did it because I wanted to access gpd4 chat because I was doing more advanced stuff um the big one of the big benefits from Chachi B4 is that it has a lot better answers lemme let me rephrase that has a much better answer has slightly improved answers um it understands the context a little more for 90 of the people.
I think using the free version is perfectly fine now if you really want consistency there and when you're using the free version there are some data caps there if they have a high volume influx you won't be able to access them that's a problem but no it is necessary. i will use it enough if you use it once or twice a day i personally use it multiple times a day for different things and it has been very worth it so yeah i guess i dont think its necessary but its helpful if you are 20 extra dollars a month and he uses it often great honestly great answer um and another question is coming up and I think this is a question that is coming up specifically from people who are thinking about um you know transitioning to data analytics um jumping into the field um do you think um chat GPT will replace data analyst that's also a question that's come up a lot yeah this is a question I've been asking myself a lot recently um and I take this question very seriously.
I have by no means taken this question lightly because it affects a lot of people and it's not just about data analytics, it's about data science, data engineering, database developers, it's going to affect all of the software users out there. some genuine way. take this question very seriously. I have a great community and I want to do the right thing with them and give them a good answer. I've been researching this, um, I think more than the average person, uh, and here's my take for today as of April 18, 2023. As of now, I've used enough to understand a lot of its limitations.
I have done enough for research to understand its limitations. I think there are some very genuine limitations on what can and will be. able to do in the future um and I think at least I'm pretty sure that the data that will be generated from these systems will be more usable because these systems I believe and again this is my personal belief based on my research is that it actually goes to make people many more companies more data centric because they have to be not because they want to but because they will be left in the dust if they don't so millions of companies that have never had a data analytics team now they're competing for data analysts who can come in and use their data because they've never done it before.
I think the larger companies that have the infrastructure to really use them well will do so less over the next 10 to 15 years, as my estimate again is hard to say how quickly things have improved. I don't think you'll see any real decrease or automation in jobs just because of a lot of ethical issues. data sensitivity, data protection issues, I think, eventually, there will be government regulations, a lot of governments around the world on the use of this and within, a lot of databases, public health information in there again, I've dipped in this too. a lot to go into in all aspects but there's something going on a lot more than I think what people think when I think automation is like it can do its job my general genuine response is right now it can't and even if could, I think there will be many more problems that it can only do its job, but how well.
Can do it? Is it legit? What are the ethical ramifications? There are a lot of things to consider, so again, I'm trying to give you the best answer I can without getting too confident. I don't think data analysis will be automated, in fact I think it will be to some extent and maybe for a short time, but even to some extent I think it will actually be more in demand than before, that's what I feel now , you can take it however you want. I work in data analysis. I am a big advocate of data analytics. Everyone knows that and you can take it however you want, but that's my genuine belief. um as of today and it's been in the past uh probably about a month since I really started to dive into that that's a great response and I think it also aligns with what we also invite other guests to talk about a Foundry career a doctor mayor I think also subscribes to that being as well and for those watching too Dr Gamera is doing an intro to data analysis workshop tomorrow I'm just going to blatantly plug that in there um I'm just going to pick up on the regulation point you made Alex because we Cynthia asked me a great question on LinkedIn about what restrictions might be put in place regarding open AI in the future.
I think I don't even know that many companies and even countries are starting to completely ban it they are having major data leaks they are having data uh misinformation they are having these hallucinations that are causing problems I am there are so many positives and I deagain before we talk a little more the positives for AI are huge which has a lot of implications for a lot of different things um there are a lot of pros but also some cons um and some of those cons especially if you look at companies at the enterprise level company that use this and where they don't have specific safeguards in place, there's a lot there could be data leaks, there could be, um, you know you're opening up your systems to really advanced software and sometimes even people chat gbt or open AI they're like we don't know exactly how this works there are countries like Italy as well as a few other countries are starting to ban or are looking to ban it entirely because of this exact issue and so companies are doing the exact same thing that I've seen.
I have read many companies that I read about. a lot of bigger companies are like okay there's a lot of pros here but now we're starting to realize all the potential cons of integrating it into our systems that we're not really sure if it's worth it because of the potential lawsuits, the potential issues of data um the potential misinformation just within your company um so yeah there will be I'm talking specifically about the US government they are so slow to act I think it will be several years before we have regulatory systems real implemented federally um but certain countries are already starting to enact restrictions um you know again there's a lot of pros but there's actually a lot of cons as well and most people like myself when i first started using it I thought there was no downside this is amazing but I consult with a lot of big companies that I work for I have worked with big companies and I have talked to a lot of industry experts who work at these big companies and a lot of them have said the same thing , they're managers, they're even entry-levels that I've talked to and they say yes, our company has decided not to use it.
I'm like, oh, don't use it. not at all they say no we ban it on all company computers you can't have it so I personally know someone who has already done it um and I think it's just a matter of time before people start trusting or not. and there are so many implications, um again, I'm trying not to be too specific, but there are a lot of positive and negative implications of using it in a company, so companies have to be very careful about how they integrate it, how they use it. if it's really useful with your current data infrastructure um and Business Systems set up in place and data pipelines as you know it's a lot more it's not as easy as people make it out to even Deploy is extremely complex um on the infrastructure real data I think that's a great answer and also reading between the lines of what you've been looking at tonight Alex um I think it's worth it before people jump into using GPT chat as well to read about the topic to see the blog posts, um, you know, check out the Press read about the implications of GPT chat um while simultaneously using it and I think that will put you in a much better position to understand not only um, you already know how it works, but also some of the security issues and maybe some of the ethical issues as well, so I think that's a great answer, Alex.
I love when we invite you. We always have your audience on YouTube and they love data cleansing. They are obsessed with cleaning data. I find it fantastic. We love to see that. um we love data cleansing huh crowd there's room for everyone but there's a big question on youtube is there specific pointers for data cleansing how can you chat GPT to help with data cleansing previous video like some ways you can use Chachi in between data analysis and i showed that data cleaning was a very simple message um do you know how i can clean this data?
That is my example of what he is giving. there is no context, you are not giving him information, more information you give him now. gbt on anyone so i'm not like a gbt chat a list or whatever you want to call me um i just really like it so i don't like the prompts developed for this but here's what i'll say um i've used it i've used it multiple times for actual data cleaning um for some projects I was doing and what I personally would do is say okay you're a lead data analyst here's the data you're working on with you are also an expert in I think that I was working with um oh what is that I'm trying to think that there's a I'll say healthcare data but it wasn't like it was something else I was working with but I was like I was an expert healthcare data analyst medical.
I thought here is the data you were given, but you ran into some issues like formatting and I would try to point out that your formatting seems to be off on this specific column here. the format I want it in what I would do is generate the code from all these different formats that were there and say this is how you can generate the code to actually fix it and then I would try if it wouldn't work exactly right. I will say that this format does not work with the code you generated. Can you also fix this guy?
So I don't have a heads up but the more information you give it the better it does and especially with data cleaning I like to be. I've been very specific, like this column. I think it should be formatted like this. How can I do that? and it works quite well. It does a pretty good job. cleanup crowd um I feel like we should do maybe a data cleansing-only live event in the future I think I'll do a whole data cleansing webinar Me Data cleansing is how I did it at my job for three years straight like this that i'm a huge fan amazing amazing um another question coming up on youtube and i guess with everything we've talked about tonight um alex in your opinion is were i going to say now is the right time to start thinking about a career as a analytics data in light of everything we've talked about tonight, yeah, another very good question, um, so I'm going to take you back to five years ago or not. getting older six years ago, when I first became a data analyst, you know, when I was there, there wasn't that much, I mean, data analytics was still very popular, but there's nothing that I believe in the last few years about probably the last two or three years analytics has gotten more popular because data science has gotten a little bit less popular if that makes sense then data analytics has had this big boom at the entry level , there's still a lot of people, there's a lot of people trying to get into Tech um, just with that six years ago it was easier, there wasn't as much competition as now because of remote work and all these other factors, so which is now generally harder to get into, but I really think within the next couple of years we'll see a lot more jobs open up, especially at the entry level with people who know tools like using GPT chat or can build more advanced projects, um so if i was trying to get in today i would say you know there is and i almost don't want to say i'm over this but i went through it about three months ago i went through this like i don't want to say like existential crisis of like my community but i was very I was worried I was very worried I was like what implications does this have for my community my data analyst Community I like this will be how I was talking this will disappear is this like it will not be one more thing The more I have used it, the more I have investigated it, the more I I realize you know there's still a lot of opportunity here so if I was starting today I'd just incorporate the BT chapter into my learning or AI tools into my learning after I learned the basics or even the stuff in between because you really can't do the job of a data analyst without at least knowing the concepts very well then you can use chat gbt because now you understand how these things work how they integrate how they work together on top of that if you can get domain knowledge i think domain knowledge it will be even more important in the future my experience as a healthcare professional, uh, you know, I think it was helpful when I first became a healthcare analyst.
It helped me get my first job in the future. mastery might even be a little more important than knowing all the technical skills because in one year two years it might be a little easier to use those technical skills even though you're at the basic level so no I don't I don't want to do it again . I've really been thinking a lot about this. I really don't think this is something people should be like, okay Jack, here the AI has taken over. I do not do it. Data analysis is going away. I don't need to be doing more.
I really think a lot of people have thought that about a lot of things like software engineering, data science, etc. the AI space I don't think so I don't think so and actually let me make one more note sorry I could talk about this for a long time one more note is you should also think about this from a business perspective when you have a big amount of data you don't want managers you don't want um you know exact C-level people doing data analysis work or asking these questions and getting AI written answers if you really think about it, even from a small business perspective, you want someone who knows this data who understands the business value of this data can interpret it and bring it to senior management stakeholders lets say data analysis be fully automated tomorrow like me i became a robot and they can hire me for ten bucks an hour let's say it happens they're still not going to want to go to the alex robot and they have to manually request it the questions mainly ask to make all these things work through issues with the AI side they're going to have someone who It's going to do that, um, I just me, even that's in a perfect world in, say, 20 30 years, it's really hard to tell. its limitations know a lot of the things i've seen issues i've seen and read about in a bunch of communities and forums i've been researching it's really really hard for me to believe that's going to happen anytime soon um or not we'll need data analysts big gig um too Alex has been talking a lot about his community and for anyone watching Career Foundry check out Alex's YouTube channel Alex Analysts on YouTube Alex is approaching 500K subscribers uh with your help tonight we might as well get a little closer while we're here and because the Foundry race team on YouTube is in the background I'll also collect the Foundry race channel on YouTube there's some great content there. data analysis too and also Alex reviewed the racing foundry data analysis program a couple of months ago so check it out if you want to dig deeper into what the program offers.
Alex, we just talked about the past, let's look to the future. You see this going in five years, ten years I know it's changing very fast, but where do you see this going in the five to ten year range? I'm going to give you a comparison when cryptocurrency really became huge, you know cryptocurrency people thought it was going to change the world. I could see what is going to happen. It hasn't exactly happened with AI. It's a little different. I've gotten so immersed in this AI stuff. um. I see the AI is getting better, um, I see it's more efficient, especially since Microsoft has picked it up, um, you know they're spending an incredible amount of money training these models to be accurate and, um, you know, uh, Sam Altman , who is the CEO of open AI, he even said himself that it is not sustainable to train with so many parameters, uh, you know we have to find better ways to do it, so they will condense it, they will do it more efficiently, which is good to the end user to us in the end because uh that just means these models will have faster iterations we'll see more improvements um yo there is uh I'm going to talk a little bit about the negative that I look at and then I'm going to talk on the positive the negative I see is I don't see the hallucinations going away actually uh open AI is open openly admitted they don't know how to solve this problem um they have. tried to tune their models um so that the data is more accurate it has to come from a certain source even then they are seeing almost the same level of hallucinations that is very real and needs to be considered.
Don't know. how to say that but it has to be kept in mind when you're using it because even if you go further in five ten years, um, we don't know if these hallucinations will ever go away and that's a problem that's part of that piece of automation that's concerning , if you have a data analyst telling you facts that aren't facts, that's always a problem, um, that's a negative that I see, um, the positive that I see is that I see these models getting better, more accurate , more efficient. um I see them embedded in a lot more stuff a lot more stuff I almost see AI being embedded in almost every aspect of the internet um ya and that's like my deep dive into things I see it embedded in all of these different products and it isjust like a person on the internet doing it, but I think you give it a year and companies are going to do that like you know it's moving, but it's not as fast as I perceive it to be.
I think it's three five ten years. far where most companies, most online um platforms will have AI in some way as part of their platform so just understanding artificial intelligence understanding these big language models that are going to change over time um it's important one more thing i want to mention is it's changing so fast if you can't keep up you're not the only one who can't keep up and i am i always look everyday i'm seeing new stuff um and i can't keep up rhythm there is also a lot of what i will say is you know the foundation starts with gbt chat like what we saw here online which to me is like the basic building blocks that i am now personally.
I just started using um uh Auto GPT if you have ever been running that on my local machine yesterday running agents where they can automatically exit and do tasks and they have a very low success rate right now like 10 15 but they know they're going to get better so these these automators um these task automators or you know Automation in general is going to get better I just genuinely see I can see the Horizon where these systems are not going to be perfect and they just aren't going to be I think they're going to change the world but I see a lot of problems with the actual quality of the products being released.
I see there are a lot of limitations, and as you use it more, I hope you'll see that it's not perfect, too. system definitely definitely trying to link some of the questions that come up on big marker um LinkedIn and YouTube so I'm just trying to link something but one question that comes up is and I know a lot of people watching tonight are thinking about taking your first steps in data analysis, um Alex, in your experience for a junior data analyst or someone who is ready to break into the industry, what are the types of tools that you would like to see given to that candidate? a Competitive Advantage, are you talking about most AI tools and only the two, data analysis type of tools, okay um so if you're just going in I still think SQL is number one that you need to learn even above huh even above Chachi BT uh you know SQL is just one of the fundamental skills you need to know highly highly highly recommend learning it getting really good at it building projects on it um it has a very low skill cap but has an extreme or so. you have a low skill cap to start such a low barrier to entry but you have an extremely high skill cap some of the best data analysts data engineers in the world are still learning things about SQL and have been using it for 30 years so , um SQL number one, then your classic Excel, you have to know how to use it and then, um, I always recommend people start with Tableau power bi for visualization because they know they're really important in the community. go to a company sometimes they won't use it they will use some other tool or they will use a tool that is inside their um that they built themselves or some 3rd party tool usually has a lot of the same features and you can figure it out very quickly if you know Tablo or power bi um and then, I myself am a huge proponent of python.
I use it a lot, um, especially for data analysis. It has so many libraries and packages that are great for data analysis. and those are generally the ones I would start with and then after that, um, you know, look, look at the cloud platforms, um, Azure is going to be Azure already big, but it's going to be bigger with AI. I'm digging into it, but AWS is always great for learning. Personally, I do not recommend the Google Cloud platform. Not the best in my opinion, but organizations use it for their use cases, so you may need to learn it and then learn AI tools. true, it's me.
I just posted on LinkedIn last week. Me or maybe it's like it was yesterday. I don't even remember that time flies in my world, so I posted. learn the fundamentals don't just go straight to gbt chat but once you learn them start learning how you can integrate gbt chat or AI tools into your, uh, you know, kind of your tool belt amazing fantastic and also how junior data analyst or someone who is Breaking into the industry What is the importance of a portfolio when you go for job interviews? Sure, maybe yes, I love portfolios that helped me land a few jobs. i don't have them which is yes the portfolio is good for two things one is good for helping you get an interview so you can send your portfolio on your resume as a link in your contact and the portfolio is going to have projects going to have like tangible projects that a hiring manager or a recruiter can see, they can see them, they can be like, oh, this guy has like four different projects on Tableau, we needed someone who knows Tableau, they look at their projects and say they're pretty. well it's another one that helps you get the interview better than that, although it's extremely helpful for people who don't have experience because you don't want to go in there and they'll ask if you know how you have SQL in your job and you're like, oh, not really i have experience using sql in my job because i have never been a data analyst terrible answer thats not what you want thats why you have a portfolio where you have sql projects um thats just a tool but it has projects in sql it can say I actually just built this project where I know I've taken this dataset and done these things in SQL and use it I know I've worked with it with data cleansing and exploratory data analysis and you know automation with store procedures and triggers and all these things and you can say you can talk to it instead of saying I don't have any experience you can say I have experience building these projects.
I think that's by far the main reason to have them will help you as well, it will help you appear a lot more believable and look more believable as well as give you something to talk about when asked about your experience in these skills not the other thing. . it doesn't mean if i take some courses on sql you need to take courses on sql do it but it doesn't mean in an interview it means well i built these projects and here i use case statements i know unions i know unions i know window functions i know stored procedures that's what I've used and then they say oh wow you should hire this guy because he knows he has the skills we need in SQL so um you know that's all is all how you frame things in interviews and you don't want to frame it that you don't know what you're doing even though you don't have experience you have experience if you've built projects that's your experience until you get a paying job fantastic experience and I would also say at this point that um just to go back to Career Foundry but the Foundry career data analysis program , the key takeaway is the portfolio, so you work through the program on a portfolio and the The best thing, as Alex just said, is that when you go to that hiring meeting, you'll be able to talk about the projects that you've worked on and explain exactly what it did step by step and it works as Alex said its a great great guide have a couple more questions via alex on youtube but also on uh great bookmark um you know where you keep up to date on all this AI info o specifically do you know data analysis in general how do you keep up to date? to date with the industry, um, okay, I'm in a lot of places, I, I, I also have a great community that sends me a lot of stuff, so, um, personally, I get a lot of stuff sent to me via email, LinkedIn, Twitter , almost every day.
I can't, it's hard to keep up with everything, but myself, if I'm just scrolling, I actually found that Twitter is the best place for AI content, um, and I just follow a lot of people in AI who are researchers in Google. and Microsoft who release updates, as well as people who are doing open source projects in AI. I'm a big believer in open source stuff. these developments that are happening live instead of keeping it behind a wall so I follow a lot of GitHub repositories as well as a lot of people on Twitter posting these GitHub repositories then I go and follow them but personally I've found that Twitter it's the best source of AI content in my opinion amazing awesome yeah there's a ton of free content great blogs I'm aware at the time Alex I don't want to take up so much of your time it was a fantastic presentation this evening.
Thank you very much for the questions from everyone in the audience. It's great to see all the interaction on Big Marker on LinkedIn on YouTube. Thank you so much for the Alex crowd. In terms of free content, as Alex mentioned, I'm just posting on a big bookmark a link to the Career Foundry short course, so if you're interested in data analysis and want to do a free five day short course and check it out and also check out check out the foundry career blog we have a lot of editors working in the background writing some great blog articles on data analytics um also some more general articles on career change and also check out the foundry career blog and also while I'm here let me change a slide we're currently offering a Foundry major scholarships to change majors so if you listen to Alex tonight you've been inspired by the GPT chat and you've been inspired by data analytics and you are you thinking about you know jumping into the industry we are offering a career change scholarship from our data analysis program for the first 50 career changes if you are watching alex's youtube channel if you just scroll down you can book a call with the advisor from the show, you can talk about the career change scholarship, but also if it would be a good fit for a career in data analytics, um, and that's about it from me, Alex, thank you so much for joining us tonight.
I'd love to let you share that candid knowledge and all your different insights into the GPT chat and data analysis industry. Always a pleasure to have you on the channel and for everyone watching um who hasn't yet please subscribe to Alex's YouTube. channel Alex the analyst on YouTube there's some fantastic content there um Alex is a rock star in the world of data analysis and check out Alex's YouTube channel Alex that's it thank you so much for joining us this tonight and we're going to I'll see you I think you'll be back on the channel soon so check out the career Foundry events page for all upcoming events and also join us tomorrow where Dr.
Humera will be holding an Introductory Workshop to data analysis skills where we will take some raw data and bring it directly to visualization and presentation. and yes thanks for joining everyone tonight and we'll see you again next time
If you have any copyright issue, please Contact