ChatGPT + Noteable (Jupyter) = Mind-blowing!

hey there python trainer ruven learner here and today I want to show you one of the most amazing amazing technological advances I've seen in a very long time I'm talking of course about Chachi BT I know you're thinking I'm a little late to this game I'm recording this in May 2023 Chachi BT has been around for an awfully long time for like six months already which I guess is not really an awfully long time but in computer years it is anyway I want to walk you through how Chachi PT can be used to retrieve data create Jupiter notebooks and even produce queries and plots now I wanted to set this up so it's going to take a little bit of time for me to just show you what's going on and uh whereas I usually don't edit my videos very much I try to walk you through what I'm doing and show it to you in real time this time I am going to actually do some editing and that's just because some of these things take a long time so let's step back for a moment to the beginning I am assuming you know about Chachi BT everyone's been talking about it for about six months now and if you go to the Chachi BT site at chat.openai.com and you have a free login you will be presented with the GPT 3.5 model and as we see if I hover over it it says this is our fastest model great for most everyday tasks and that is true GPT 3.5 is a fine model for doing an awful lot of stuff and for the last few months they've been offering what they call chatgpt plus a paid plan where you pay twenty dollars a month and then you get faster access you won't get those annoying messages that the system is overburdened right now fine I was actually kind of cheap and I decided not to go for that um until I discovered that if you sign up for chat GPT plus you get access to another model a second model gpt4 as it says here if you hover over it our most capable model great for tasks that require creativity and advanced reasoning there is no doubt the gpt4 is smarter and better other people have written and shown stuff about that but there are two pieces to this that I think are interesting as well and which Chained and just changed in just the last few days if you go down here you can see it says browsing and plugins plugins is where it's really at plugins are well I mean we know what plugins are for all sorts of software systems right there are now plugins for gpt4 and if I click on that well we'll see here a little menu I've already gone through the process of choosing plugins but let's take a look at what it looks like you get this menu here now if I go down here to the plug-in store it will show me all the different plugins that are currently available for chatgpt plus users when they're using the gpt4 model so you have to be a paid user and you have to be on gpd4 and then you can install any or all of these plugins so if I look at you know popular ones very interesting nice zapier I'm not doing this video about zapier but Boyle boy you can do amazing stuff about zapier on and on lots of different things that you can install you can look through the new ones only you look at through all of them and go through them and just a few days ago when I started using these plugins I think they were 10 pages now there's 16 pages so clearly a lot of people are adding plugins one plugin that I want to show you is notable and I'm going to go to the notable site in just a moment I just want to show you when you want to install plugin you click on install if you want to uninstall then you click on uninstall that's pretty intuitive well they don't tell you at any point is that you can only have three plugins installed and activated at any time so I when I installed a whole bunch of different plugins but then here up at the top it says three out of three are enabled so I have installed and enabled notable which I'll tell you more about in a moment world news because it will give me access to news headlines we'll look at that moment and zapier just because oh my God it's so cool again that's the time for another video so what is this notable now I want to tell you I never ever heard of notable before aside from knowing that some famous people in the python Community are associated with it until about a month ago when it's at pycon us in Salt Lake City I saw their Booth I went up to them I chat with people there I actually have a meeting with them in a few days to talk about maybe using their product um I am a free user of the product I'm not paying and everything I'm going to do here is just as a free user from the outside so what is notable if you go to notable.io and that's notable with n-o-t-e notable it's a company that's offering data analysis with jupyter notebooks and you might be saying Jupiter notebooks right like I've been using Jupiter notebooks for a while that's true you can think of notable as providing us with Jupiter notebooks on steroids so if I go to get started I guess I have to like go to my login here it's a free sign up I signed up for free you then have to create let's see what we got here here we go so it's going to show me my space so this is like not to be confused with Myspace that's gonna allow a way a long time ago so this is Reuben space here okay I'm gonna not gonna leave a review and Reuben space because I logged in and I signed up before again it's a free sign up you just have to then confirm with your email and I created a project and you think of a project as like a directory in which you're going to put all of your Jupiter notebooks that I'm going to need and I called mine chat GPT playground and this is actually important in the URL you'll see it says app notable i o p and it has this long list of hex digits here and then the name of your project we're gonna need this I'm going to keep this tab open we're going to come back to in just a moment so what we have is Chachi pt4 the gpd4 model I'm in chatgpt plus which gives me access to the four model I have the notable plugin installed and I have it also activated so what what can I do with this now I'm going to warn you right now I mentioned before right this is gonna I'm gonna do some editing here uh maybe I didn't mention maybe I edited that out right uh even I'm going to be editing this video because Chachi pt4 is very very Advanced and very slow so I'm going to show you some parts of what happens when we're using the model but then I'm going to fast forward so you don't have to worry about like sitting through it's like watching grass grow to some degree but jbd4 with these models how about this I'm going to say here um show me 10 headlines from recent newspapers about traffic accidents in the US and uh also a publicly accessible and let's say this for every headline show me a publicly accessible data source using you know in CSV or Excel format that I can download and use to analyze traffic analyze traffic accidents show me the headlines and sources in a table with two columns now this is a sort of sophisticated query that you can do to chat GPT with a gpt4 model now when it sees that I want headlines you can see it's thinking for a while here when it sees that I want headlines it's going to be like oh wait headlines if I'm going to use headlines I'd better go use a plug-in and look at this it says using world news so it first had to interpret what I was asking it that's a prompt by the way people are talking all the time now about prompt engineering first I interpret what I was saying it goes it asks world news and it says here are the top headlines about traffic accidents in the U.S and it's going to go and like give me this information here all right and it's going to give me the links but it's not really giving me links to data sources is it no it's not so I'm going to say well I'm going to give it a chance to do a little bit more but then I'm going to tell to stop generating now there is a danger here and the danger here is that when you are using gpt4 you only get 25 queries every three hours so you've got to be very thoughtful and careful about what you're doing so I'm going to rewrite this I'm going to say um create a two column table in which the first column contains let's say a silicone table let's just say with five rows in which the First Column contains a headline about traffic accidents or problems in the United States from U.S newspapers and in which the second column contains a link to a publicly accessible CSV or Excel data set let's say just your CSV data set relevant to the article you brought up let's see if this does it a little better all right now notable why is it using notable well actually it's because I asked for a data set so actually it's going to give me some trouble here I'm going to tell it to stop generating and by the way we can go look what did it send to world news and you can see I'm just going to scroll up here it said traffic accidents in the U.S so these are the responses I got back from world news and you can see it's all in Json format so we can see sort of behind the scenes what's going on and notable it's saying here okay I'm going to create a new notebook and you can already then see some interesting stuff here but this is like really not what I wanted so let's try this once again one last time before I go back to uh prompt that I've used in the past create two column table with five rows which is the first column about track of accents and which the second column contains a link to a publicly accessible uh data set okay do not use notable or create a Jupiter notebook rather just show in the second column a link to the C SV file that I can download if I wish okay let's see how it does now sure here are the headlines and corresponding data sources okay look what it's doing now now it's finding me headlines now this already is extraordinary right this is the sort of research that grad students have done for their advisors for years and years and so what it's doing is finding right headlines and it's finding a data source and it's giving me a link to that data source already pretty darn amazing okay here we go and we've got our five rows and it gave it to me you may need to sign a book waste request access okay that's fine now if I hover over this we're going to see that this is like data sets us accidents fatality analysis by the way I know this is kind of morbid but fine so let's now see if we can use this to create a Jupiter notebook so here's what I'm going to do I'm going to say using the data at and I'm going to just copy actually let's go into a tab here let's just see what we got and here you can see there is CSV here that we can download this data is provided in terms of the CSV file let's see if I can download it somewhere ah click on download you're going to use this okay fine so I'm going to download the data frame here so I'm just going to copy that actually I'm just gonna click on this and see what we've got oh now we've got to download it so that's probably a little big 200 megabytes to force into this so let's go back and look at the next thing all right here let's go here this is why I always like to get multiples Okay so we've got here our mission oh boy oh boy oh boy okay let's see if I can just go into Fred I'm gonna I'm gonna shortcut this here so first of all I think it's pretty darn amazing that we can get this information back in chat gbt it is pretty amazing so let's look at traffic accidents and the reason I'm using Fred is just because I don't really care like I don't care I don't know why I'm so stuck on traffic accidents I just want to get data here and Fred from the St Louis fed gives me a lot of data so I've got data here right I didn't get it from my chart fine I'm gonna fire that research assistant 20 bucks a month I can get another one for a same price but here I've got information about Freight rail Intermodal traffic do I really know what that is no am I interested in it quite possibly here's the thing Fred can give me a lot of interesting information so here's what I'm going to do I'm going to find this and I'm going to find here the federal funds rate which is like the interest rate that you have to deal with that's what we talk about we talk about uh interest rates that the FED is setting the Federal Reserve in the U.S so here I've got this data set here I've got this data set so here's what I'm going to do so using so download the data from and I'm going to give it this URL sometimes it's okay with figuring out the URLs on its own sometimes it's not and turn it into a Jupiter notebook in uh notable all right then I wanted to get the other one here download this hmm all right then download uh information from here and put it into a second Jupiter notebook in notable then now why would I want to do both of these right so this is going to show me the traffic that's going to show me the federal funds right then join these two data frames together and a plot it create a line plot using Seaborn showing uh both of these data sets across the years this is where the magic is gonna happen friends because what is it going to do it's going to download the data from one it's going to download the data from the second it's going to join them together and it's going to know to join them together on the years and then it's going to do a line plot well at least that's the plan now there are going to be or there possibly going to be some problems with this problem number one is that it often doesn't know which space I want to use in chat gbt I'm sorry in notable so if it gives me problems like that then I'll need to go in here and I'll need to copy this part of the URL and tell Chachi BT hey I want you to learn I want you to use this and you have to give it that you can't give it the name of the uh play the space you want to use of the project you want to use rather you have to give it that set of hex codes now what is it doing with notable what is it sending well it's saying here we're going to create a new notebook so create a new notebook name okay and it says here the user can follow along with your cell that was in the first one and the second when it says we're going to download this data see I didn't have to tell I mean it says import pandas SPD again this is all in Json so we can sort of kind of look at what it's doing here it's saying that it wants to download all this information then what does it do it's going to download into oh it's into a separate one oh boy okay got to be really good at this stuff here all right it might have been able to join them together but I'm a little skeptical so I'm gonna just stop it there I'm gonna say here I'm going to say download into a Jupiter notebook then use the set then download into oh I said a second Jupiter notebook you know what the problem is not chattybt the problem is me into the same Jupiter notebook but a different uh pandas data frame my bad all right now let's Let It Go see now people are saying that maybe just maybe uh we won't need programmers anymore because people will be able to describe what they want to chat gbt and then programs will be obsolete but as you can see you have to specify it really really carefully to know what's going on so what's it doing now it's sending a request right and saying I want to download from this URL you can see here's the URL it's going to read the CSV it's going to set this here we go and then it's going to look at this convert the date column to date time format it's doing this in both of my data frames would you know to do that maybe but it knows how to do that then it sets the index for both of them to be the date then it joins them together and the response is all this stuff together the kernel is ready and then what is it going to do and then it's going to import import Seaborn and it's going to run it and in just a few moments we are actually going to be able to compare these here is the line plot it's showing me the line plot and I can access it now how is it that it knew where to actually plot things well uh how does it know like how does it know which jupyter notebook to use because I'd already set my project in the past again if you need to set the project you just get those hex codes from there now I have to say these two pieces of information are completely irrelevant to one another and yet and yet this is pretty pretty darn amazing as far as I'm concerned I gave it some text I said download data from here download data from there join them together and produce a line plot and I said I wanted to have a line plot in this way and that's what it did now this is very nice but can I see the jupyter notebook that it did absolutely I can we go in here less than a minute ago this was changed and you can see here's the URL to download this from here's the row downloaded that from here is where it did all of the pandas stuff and here's where imported Seaborn wait a second wait a second oh yeah here we go that's the next line but for a moment I thought it was lying to me and not doing it in the correct way now what if I want to do some other kind of plot right I mean what what if I want to I mean there's so so so many ways that we can do things and because I told to do a line plot that was fine what if I say now create a histogram of fed funds rates um so that I can see you know yeah over time well let's just say fed fund rates let's just do that keep it nice and simple it's going to know or it should know which of these data frames it should pull things out of and how to create a histogram now what is it doing when I say use notable look at this it's going it's thinking it's doing and here's the histogram of the fed's fund rate look at that it did it and now if I go back into my Jupiter notebook you will see it added it here in the background and you'll also see it used SNS his plot in other words it knows that I prefer to use Seabourn and thus it's going to use Seabourn for that now I'm going to stop here but from my perspective this is one of the most extraordinary amazing things ever I can basically go into Chachi PT now I can ask it to find me headlines and data sources assuming that those data sources are actually CSV files and not just links that I have to then well obviously I like yeah I'll give that examples here you can then give it those URLs it will download them and it will then incorporate them in fact watch this what if I now I'm just going to try one last thing here because you never know oh no this is like too big two gigabytes of travel information I'm not doing that all right so not only has this been a fantastic advertisement I guess for Chachi PT Plus and for notable but it is just mind-blowing to me now what we can do from within chat GPT how we can describe things how it takes what we say literally how we're able to create plots and graphs and how we can modify what we want and then get amazing results this is not the last uh video that I'm going to be doing I just wanted to get this out there because my mind has been blown so much over the last few days please if you have suggestions comments uh thoughts put them in the comments I want to hear from you I'm going to be experimenting with this a lot I'm also going to be using it in this week's bamboo weekly um because like how can I not uh bamboo weekly if you're not aware is my newsletter on sub stack where I take topics from current events I analyze them using pandas and I'm always trying to find new functionality in pandas as well as interesting things in current events if this isn't new functionality that we should be playing with I don't know what is folks I hope to hear from you I'll be back soon with another video probably more chat GPT stuff talk to you soon hope to see from you in the comments

Channel: Python and Pandas with Reuven Lerner

Views: 155,784

Keywords: python, python3, python standard library, learn python

Id: otk2AA6hGgU

Length: 20min 19sec (1219 seconds)

Published: Sun May 21 2023