ChatGPT for Multiple Regression

all right good day everyone today we are going to explore chair GPT and how we can use it to write some machine learning code for us chat GPT is a language based model that has been created on 175 billion parameters and is released by open Ai and it's free for everyone to use let's take a look can you perform some Eda on fuel consumption Dot CS3 which is a data set I've obtained from kaggle you are free to use any data set of your liking in r just ask a very simple question let's see how it does for me and it remembers the context of the chat so every time you ask it to do something else it will get more and more intelligent and I've asked you to do fairly complex tasks well I I code and only gets better all you need to do is probably understand the correct interface because you need to see how you can make the code better and you can give it feedback and it will really respond to the feedback very well and make your code more better less verbose and highly efficient as long as you know what you need to ask it it's extremely clear you just need to remember your Concepts and no syntax at all you can see just generated my GG plots my summary statistics the structure of the data set is creating Scatter Plots so pretty good so let's see if it works as this or if I do need to make some modifications I will ask it to make the modifications itself as uh I want to explore how intelligent it can get you can edit it yourself so but I want to explore on how charging Beauty works on its own so fairly intuitive you can see all the plot is coming all the plots are opening on the same window I want to ask you to open on separate windows so let me see if I ask it can you add X11 which is the command to open the plots on separate windows in our in the above four to open the plots in separate Windows now as I mentioned it learns of the context so let's see if it's intelligent enough now to add it at the right place so that I need not remember where to put it or I need not focus on where I need to handle in which section so let's see it's writing the same code for me again and just added X11 at the right place just before we open the plot and similarly is done for the other block system so fairly intuitive fairly intelligent and I would say extremely useful and mixed extremely helpful to exploit this period which you develop code and you can also ask it a fairly complex tasks for you as long as your conception knowledge is pretty strong so it's a great utility it's done is just modify the I could have done this myself in the previous code but I wanted to show that you don't need to know where you need to add it at the right place as long as you know what you can ask it you could probably even ask it what is the function name to open it separate with those and it would tell you you don't need to remember that as well so pretty good let me copy paste and see this here yes you can see multiple windows are opening now you have box plots you have Scatter Plots different variables I could have had it to a pair plot but the purpose is not to get into the details of the exported data analysis or you know we have histograms but just to understand how this can be extremely useful for you if you want videos to explain a full machine learning project using this please to let me know in the comments and I will make one the purpose of this video is to just to explore how chat GPT can help you write code without you having to write or edit a single line of code so that we ask a two or multiple regressions so let's see can you perform a multiple regression on the above unit now I have not given it which is the independent variable which is the which are the dependent variables let's see all 400 figures on its own and then you can obviously modify that or ask it to modify that as per your liking what is taking fuel consumption consumption or combination as the Target variable address of the and the rest of the variables are the dependent variables I would like CO2 emissions so better so let's see uh if I can ask you to add can you modify the code to use CO2 emissions as the independent variable let's see so hopefully I would like to predict the CO2 emissions so let's see it has done that obviously in the previous comment I'm not given it any kind of whatsoever as I mentioned the better you get it yes it has added CO2 emissions now you can see and just taking the other it's been a best case on the other numeric uh dependent variables but as I mentioned as long as you guide it just get more and more efficient and improves all the time so let's see how this does [Music] and you can see it has done a great job it's already different the model let's take a look at the model so let's show the residuals a standard error and the adjusted R square is 0.86 right so it's very very good that means 86 percent of uh uh I mean the independent variable can be predicted up to 86 percent by the dependent variable so it's fairly very uh you know a good model and obviously I have not done anything I have not even explored which variables I've got predictability just the purpose of this video was just to explore how it can generate the score for you if you know more stuff in machine learning you can ask it to do more and more complex uh stuff for you I've asked you to do a lot of um you know great and complex code and it has done extremely well and gets more and more interesting so the model it will also generate in the line for you so a pretty good model and it is generated this code for you without me having to write a single line of code at times you need to edit it to make it more uh better but I also wanted to explore if I could just guide it to do what I want to do without me having to edit the code as I mentioned when I opened a new window I asked her to do that for me and it inserted in the in the appropriate place so without even having to change the code you can have a tool it it you know I do for you right I could ask it to for example can you do a pair plot for me in the above course so I don't need to know why I need to put it and it will automatically add that in a appropriate way so that's about it for now if you want me to make some more videos please do like and comment below on any specific machine learning problem and I can explore that the it does not it took the purpose of this video was not into getting the depth of machine learning but just on how this would like straight your course value right so it is pretty well so all right it's uh just put the this is the pair product you see so it's pretty good so that's about it for now I'll see you later and do let me know in the comments if you want me to make some other videos as well thank you

Channel: wesley weel

Views: 2,672

Keywords: ChatGPT, Machine Learning, Multiple Regression, AI Code Generation

Id: i-Qmi-7M3N4

Length: 7min 48sec (468 seconds)

Published: Thu Dec 29 2022