WEBVTT

00:00.000 --> 00:01.320
-: Hey, welcome, and in this video,

00:01.320 --> 00:03.240
we're gonna explore how we can use ChatGPT

00:03.240 --> 00:05.790
to generate things other than just images and text.

00:05.790 --> 00:07.800
So how can ChatGPT transform your data?

00:07.800 --> 00:10.980
Well, ChatGPT is capable of not only just generating text,

00:10.980 --> 00:12.900
but also other file formats.

00:12.900 --> 00:16.080
These could include, for example, CSV files, text files,

00:16.080 --> 00:19.170
markdown files, or even .PPT files,

00:19.170 --> 00:21.240
which are Microsoft PowerPoint files.

00:21.240 --> 00:22.410
Why is this important?

00:22.410 --> 00:23.760
If you're spending less time

00:23.760 --> 00:26.550
converting the same information into a different format,

00:26.550 --> 00:28.560
that can be a great time saver at your work.

00:28.560 --> 00:30.660
Also, you can take unstructured texts

00:30.660 --> 00:33.510
and convert that into a structured output for data analysis

00:33.510 --> 00:35.160
or for another downstream task.

00:35.160 --> 00:37.860
Okay, so what are the two main ways that you use this?

00:37.860 --> 00:39.420
One is to take some texts

00:39.420 --> 00:41.610
and convert that into a structured output file.

00:41.610 --> 00:43.530
Another way is to convert files

00:43.530 --> 00:45.750
back into text or put a link in the video

00:45.750 --> 00:49.440
so you can download this social_media_posts.TXT file

00:49.440 --> 00:50.700
and then we're gonna analyze this.

00:50.700 --> 00:53.460
So if I go and open this, this is a text file

00:53.460 --> 00:56.070
and it has a bunch of different social media posts

00:56.070 --> 00:58.170
that we want to extract some structured data from,

00:58.170 --> 01:00.930
and we're gonna get it to generate a CSV file.

01:00.930 --> 01:04.770
So take all of that information, copy these to to ChatGPT,

01:04.770 --> 01:06.120
I want you to post these in,

01:06.120 --> 01:08.160
and then I want you to say,

01:08.160 --> 01:13.160
I want you to generate some structured data

01:14.400 --> 01:17.043
in the form of a CSV.

01:19.770 --> 01:23.220
For each social media post,

01:23.220 --> 01:25.840
we would like to extract

01:28.020 --> 01:33.020
topic, length, engagement rating from one to 10,

01:36.360 --> 01:41.360
10 being the most engaging and hashtags.

01:42.090 --> 01:44.465
And so what we have here is we have our data,

01:44.465 --> 01:46.410
which is in raw text,

01:46.410 --> 01:48.750
and we've told ChatGPT exactly how we want

01:48.750 --> 01:50.520
to structure this data.

01:50.520 --> 01:53.280
We want to export a CSV file.

01:53.280 --> 01:56.700
And for each of these, what we want to extract is the topic,

01:56.700 --> 01:58.860
the length, the engagement rating

01:58.860 --> 02:00.810
so we're gonna leave it up to ChatGPT to decide

02:00.810 --> 02:04.050
what the engagement rating is and the hashtags.

02:04.050 --> 02:06.600
Cool, so I'm gonna leave this to run now.

02:06.600 --> 02:08.326
ChatGPT has decided to use Pandas

02:08.326 --> 02:11.100
and it's decided to use the code execution.

02:11.100 --> 02:13.140
It's creating this data so you can see

02:13.140 --> 02:14.730
that if you click on the analyzing,

02:14.730 --> 02:16.590
it's filling all this information in.

02:16.590 --> 02:19.230
And then we'll just give it a second to see what happens.

02:19.230 --> 02:20.970
So it's figured out what the topics are,

02:20.970 --> 02:22.560
the engagement ratings,

02:22.560 --> 02:25.560
and then it's also then creating this data.

02:25.560 --> 02:28.050
And there we go, we have our interactive table

02:28.050 --> 02:29.550
that's been produced here with the topic,

02:29.550 --> 02:32.460
the length of the engagement rating and the hashtags.

02:32.460 --> 02:35.040
And we can download that if we click on the download

02:35.040 --> 02:37.200
and you can see this is the social post analysis.

02:37.200 --> 02:38.970
So we can click on that, download it,

02:38.970 --> 02:40.440
and if we have a look at the CSV,

02:40.440 --> 02:43.410
you'll see that we've now got the data in a CSV file format.

02:43.410 --> 02:45.120
So that is a great way for us

02:45.120 --> 02:47.340
to be able to easily get access

02:47.340 --> 02:50.610
to structured data directly using ChatGPT,

02:50.610 --> 02:52.800
and its code execution functionality.

02:52.800 --> 02:54.960
The next thing I want to show you is let's go back

02:54.960 --> 02:57.000
and use the students.CSE file

02:57.000 --> 02:58.710
that we had in a previous lecture.

02:58.710 --> 03:01.470
I'm gonna go and click on a new chat,

03:01.470 --> 03:04.623
and then I'm gonna add the plus from the students.csv.

03:05.610 --> 03:09.210
And then we're gonna say, can you generate a summary

03:09.210 --> 03:11.190
for this CSV file?

03:11.190 --> 03:14.490
And what we're doing here is taking structured data

03:14.490 --> 03:16.260
and we're then getting ChatGPT

03:16.260 --> 03:18.480
to create a summary for this.

03:18.480 --> 03:20.340
And here you go, so it's done some code execution

03:20.340 --> 03:21.690
on the data

03:21.690 --> 03:24.120
and then it's identified some of the key columns,

03:24.120 --> 03:28.410
the highlights, and so we can actually take structured data,

03:28.410 --> 03:30.030
like a CSV file

03:30.030 --> 03:32.160
and get directly some text

03:32.160 --> 03:34.590
that represents what's in that data.

03:34.590 --> 03:37.530
So the final thing I wanna show you is you can make this

03:37.530 --> 03:39.000
into a PowerPoint presentation

03:39.000 --> 03:41.220
as well using chat's code execution.

03:41.220 --> 03:44.730
So we can say, can you take these insights

03:44.730 --> 03:49.680
and make this into a Microsoft PowerPoint presentation

03:49.680 --> 03:54.360
so that I can easily show this with my boss?

03:54.360 --> 03:56.640
So now what we're gonna do is tell ChatGPT,

03:56.640 --> 03:58.050
now that it's got that data,

03:58.050 --> 04:02.460
it's now gonna use a different package called PPTX,

04:02.460 --> 04:04.680
which allows it to create slides.

04:04.680 --> 04:06.900
So you can see it's adding a title slide here,

04:06.900 --> 04:08.580
some general information.

04:08.580 --> 04:10.470
It's adding some key columns here,

04:10.470 --> 04:12.240
so it's actually adding some information

04:12.240 --> 04:14.880
and creating a Microsoft PowerPoint out the back of this.

04:14.880 --> 04:16.050
So we'll just wait for this to finish

04:16.050 --> 04:17.730
and then I'll show you the end result.

04:17.730 --> 04:18.960
Cool, so now that it's finished,

04:18.960 --> 04:20.810
we're gonna go and download this file

04:21.810 --> 04:23.340
and I'm gonna go and have a look at that.

04:23.340 --> 04:24.930
And here is the PowerPoint presentation.

04:24.930 --> 04:26.985
So you can see we've got like a title slide

04:26.985 --> 04:30.240
and we've got a general info.

04:30.240 --> 04:33.120
We have some key columns that it's identified,

04:33.120 --> 04:36.240
the highlights and the academic performance.

04:36.240 --> 04:38.850
Now I know what you're thinking, this isn't styled well,

04:38.850 --> 04:40.530
but we can keep going with ChatGPT,

04:40.530 --> 04:42.300
telling it to add styling,

04:42.300 --> 04:44.460
telling it to add different formatting,

04:44.460 --> 04:46.770
and we can get PowerPoint presentations made

04:46.770 --> 04:50.940
off of real data that we can easily transform from a CSV

04:50.940 --> 04:53.700
or some text into a PowerPoint presentation.

04:53.700 --> 04:57.000
So hopefully this helps you to go beyond just using ChatGPT

04:57.000 --> 04:59.040
to generate a text, thinking about lots

04:59.040 --> 05:00.630
of different output formats

05:00.630 --> 05:03.330
that you can use when you are at work or at home.

05:03.330 --> 05:05.080
Cool, I'll see you in the next one.