YT YouTube2Text

YouTube2Text → Transcripts → Data Analysis using AI Tools | Quadratic | N8N | Supply Chain

Data Analysis using AI Tools | Quadratic | N8N | Supply Chain — Transcript

by codebasics · 10,289 words · 1,502 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00We are going to build end to end
0:01analytics projects using AI tools in
0:04supply chain domain. In terms of text
0:06tech, we will use Netn which is an AI
0:09workflow automation tool to pull data
0:12from Excel files and it will ingest the
0:14data into Postgress SQL database. Then
0:17we will use another AI tool called
0:20quadratic which is an AI powered
0:22spreadsheet that will pull data from
0:24Postgress SQL database and we'll show
0:26you how you can perform analytics using
0:29some prompts in this quadratic tool.
0:31This project will teach you important
0:34supply chain domain concepts as well as
0:36it will teach you how you can build
0:38projects using the modern AI tool
0:41mindset. This project is not only uh
0:44perfect for your learning but it is
0:46something you can add in your resume as
0:48well as your project portfolio. Let's
0:50begin with some interesting story line.
0:55Atlart is a Gujarat based organic food
0:58manufacturer. They specialize only in
1:01few products but disrupted the market in
1:04the two cities they operate.
1:07In an interesting move, they opened
1:10their third city of operation in New
1:13Jersey, America, and started performing
1:16well there as well. Though they handle
1:19very less products, their supply chain
1:22is immature. Bruce Hurali, the chief
1:26operating officer of Atlique Mart, has
1:28noticed that some of their supermarket
1:31customers are dissatisfied with their
1:34order management and he wanted to fix
1:37this before scaling further. Bruce
1:40discussed this with Tony Sharma, the
1:43head of analytics at Atlake Mart and
1:46understood the problem is a classic
1:49supply chain issue where they are not
1:52able to maintain an optimum inventory.
1:54Tony offered to create a quick insights
1:57report using Excel or PowerBI, but Bruce
2:02insisted that he needs something AI
2:04powered as a solution because like any
2:08other senior leaders in the world, Bruce
2:10does not want to miss out on the AI
2:13hype. Tony assigned this work to Peter
2:16Pande who is a curious data guy in the
2:20team exploring all the AI tools in the
2:23market. Peter, who always thrives to
2:26drive the extra mile, found quadratic,
2:31which is an AI powered spreadsheet that
2:34even offers connection to their database
2:37in Postgra DB. He even figured out an
2:41automation using N8N to migrate the data
2:45directly from emails to Postgra.
2:49So you will be solving the supply chain
2:52problem along with Bruce, Tony and Peter
2:56using Quadratic, the AI powered
2:59spreadsheet N8N to automate the data
3:02migration and learn important supply
3:05chain domain experts. Let's get started
3:08folks.
3:11[Music]
3:20Let us discuss the technical
3:22architecture for this project. We will
3:24get our data through email.
3:27You'll get these kind of emails which
3:29will have files attached to it. I have
3:33two emails. One for India sales, one for
3:35USAS sales. And when I open this email,
3:38I will see this kind of CSV file where
3:41this file contains aggregate data. You
3:44can see order ID, customer ID, order
3:47placement date and few other fields. The
3:49second file contains the order line
3:53items. So order ID, order placement ID
3:56and there are some detailed fields like
3:58the product ID, the order quantity and
4:02so on. So all of these data is coming
4:06via email in your inbox. And by the way,
4:10we are going to provide you these files.
4:12So what you can do is you can take these
4:15files and you can send it to your own
4:18email and you can configure your
4:22automation workflow to monitor that
4:24email. So let's talk about the
4:26architecture. So here are the emails
4:29that you're getting and they have this
4:31uh spreadsheet attached to it and then
4:35you are using uh let me just draw one
4:38more block here. So you are using this
4:42tool N8N for the automation. Okay. And
4:47this tool is basically
4:51an AI agentic automation tool and it is
4:57monitoring
4:58these uh emails. So you can configure it
5:02to monitor your email inbox. You can
5:06provide the labels, the subject line,
5:09you can provide all that filtering.
5:10Okay. And as a next step what you're
5:12doing is you are ingesting the data uh
5:17into your posgress. So here will come
5:21your posgress database. Okay. So this is
5:26your posgress database and then you are
5:29attaching the quadratic AI. So let's say
5:33I have my
5:35quadratic tool. Uh it is your AI
5:38spreadsheets.
5:40So you are pulling data from posgress
5:45into quadratic and here you are doing
5:49your analysis your promptbased
5:52AI analysis is being done in the
5:54quadratic. So this will be your
5:56architecture. Just to summarize, emails
5:58are coming to specific inbox and this is
6:00what happens in the industry where some
6:04vendors or let's say some team member
6:06will be sending all these Excel files
6:09via email in specific email ID and inbox
6:13and you can configure your N to monitor
6:15that inbox. In nin you will do all the
6:18processing. We'll look into that in
6:19detail. And then data goes into
6:21Postgress and from Postgress you would
6:24pull the data into quadratic to do your
6:27analysis.
6:31So I'm assuming that using the link in
6:33the video description below, you have
6:36received all these files and you have
6:37sent two emails. one containing data for
6:40US, the other containing data for India
6:44into your personal inbox or whatever
6:47email id that you want to use for this
6:49project. Okay. And now we are going to
6:51set up N8N.
6:54In case if you have not used this
6:55before, this is an AI agentic workflow
6:59automation tool which is getting very
7:01popular nowadays. Go to n8.io.
7:05Sign in using whatever credentials. I
7:09have already signed in and you will get
7:1314 days free trial.
7:20So I'm on the homepage. So here when you
7:23come for the first time you will uh see
7:25little different UI. I mean all these
7:27things will still be there. So you're
7:29going to create a workflow here and in
7:32that workflow you will add a first step
7:35and this first step is to monitor
7:39your Gmail. So you click on Gmail on
7:43message received. So here whenever you
7:47receive a message into your Gmail
7:48account you want to trigger this
7:51automation workflow. These workflows can
7:53be triggered via different means. In our
7:56case, it is triggered whenever new email
7:59comes to our Gmail account. Now here I
8:02have already added my authentication.
8:04But essentially what you can do is
8:06create new credential and here sign in
8:09with Google. So you will say sign in
8:11with Google. You will provide all the
8:13credentials. It is pretty common sense
8:15folks. You should know how to do it.
8:17Okay? And if you don't just take help of
8:20chat GPT. So you will uh log to your
8:24Gmail credentials and after you have
8:27logged in you will get this thing okay
8:30Gmail account so I have configured that
8:33particular email id right this
8:35particular email id I have already
8:37configured it and then here
8:41I'm saying monitor this Gmail for every
8:44minute for incoming emails and here uh
8:49just uncheck this because we will need
8:51this option and in add filter you will
8:55say labels okay label label names or ids
9:00and I want to use my inbox so I'm
9:03monitoring my inbox essentially
9:06then search
9:09so in the subject you will say
9:12okay what is the subject look like so in
9:15the subject you will say daily sales so
9:18all these emails will have specific
9:20subject subject. So you have to identify
9:21that pattern. So I'm saying that if my
9:24subject contains daily sales. Okay. So I
9:28will say subject contain daily space
9:31sales then look at that subject and in
9:35the options you will say access the
9:38attachments. Okay. So once again uncheck
9:41this simplify only then you will see
9:43this particular options. Now I can say
9:46fetch test event and see it is able to
9:50fetch that particular file. So if you
9:52look at this particular email what it
9:54did is it went to this particular inbox
9:58and it pulled this particular email okay
10:02so you have two attachments okay 4.8 8
10:05KB 20 KB and when you look at it 4.921
10:10right so so almost same so fact
10:12aggregate India so this is the email for
10:14India so it it pulled the most latest
10:17email here so this is looking good in
10:19case if you want to view the file you
10:22can click on view and you should be good
10:24to go click on back to canvas button so
10:28this thing is good the next step here is
10:32to extract the data so we will say
10:35extract from file. So now we are pulling
10:38those files but we need to extract data
10:41in a JSON format. Okay. So we will say
10:45extract from CSV
10:47and here
10:49you will provide this field. So you are
10:51saying that we have two attachments. One
10:54is aggregate data, one is detailed order
10:56line data. So we want to firsth process
11:00this particular file which is your
11:02aggregate file. And here if you click on
11:04JSON
11:06you you see this if you look at JSON it
11:08is mapping this 128 items and it is
11:12pulling all these fields. All right. So
11:15this part looks good. Now if you just
11:18say test step it will actually show you
11:20these. Okay. So this you will be able to
11:22see when you click on test step. So
11:25that's it I think. So essentially you
11:29are converting your CSV data into JSON
11:31format through this step. The next step
11:34then is to uh get this data and ingest
11:38it into Postgress. Okay. So here
11:45click on Postgress
11:48and insert rows in table.
11:52So now you need to connect it to your
11:55Postgress account. I have already
11:57connected it with my posgress account
12:00but for you you will click on create new
12:02credentials and it will ask you for all
12:04these details. Now you need to use uh
12:08superbase. Okay. So superbase.com go to
12:11that website and create your login. You
12:13can create your free login. So I'm
12:15already logged in into this account.
12:18Superbase is a way to host your
12:20Postgress database on cloud. See you can
12:24run Postgress database on local machine
12:26but this allows you to run it on a cloud
12:29and Postgress just in case if you don't
12:31know is a relational database that also
12:34provides no SQL capabilities but here we
12:37are essentially using it as a relational
12:39database only. So create an account
12:42using I think Gmail or whatever again
12:44pretty pretty common sense uh thing that
12:47we are doing here and you will have some
12:50organization etc. Uh so create an
12:52organization if you have not created
12:54already and when you go here uh you need
12:57to create a project and by the way we
12:59are going to provide all the steps
13:02related to this in a PDF file. So once
13:05again check video description you will
13:07get a link through which you will be
13:09able to download all these files and one
13:11of the files will be to set up your
13:14posgress on superbase. Okay. So please
13:19follow those uh files we have given
13:22everything in detail. I have already set
13:24up my database because I don't want to
13:26waste uh time and I I don't want to make
13:29it like a very long tutorial. So I
13:32already uh set up my database and my
13:34database looks like this. Once again
13:36these these steps are provided in the
13:38video description. So check it out.
13:45Okay. So these are the tables that I
13:47created and you as you can see I have
13:50two fact tables and three dimension
13:53tables. Just in case if you don't know
13:56about fact and dimension table I have a
13:59video on YouTube this one star schema
14:01video you you can watch it but
14:04essentially fact table contains your
14:06transactions whereas dim table contains
14:09your dimensions such as customers you
14:11know the information that doesn't change
14:12too often customers products and all
14:15that but fact is more like a
14:17transactional data you you see it kind
14:19of changes and if you look at the data
14:23here we have a data till 56
14:26you know you can sort descending and we
14:29have data till 16 May right sort
14:32ascending let me see so order placement
14:36date sort descending 516 okay so I'm
14:41assuming you have followed all the
14:43instructions from that PDF file that we
14:45are attaching and your database is ready
14:48now to get the connection string you
14:50will go here okay and you will look into
14:53your session session puller and you will
14:55use all this information host port etc
14:58and you will enter that information
15:00here. So for example my host will be you
15:04copy from here and you you know this is
15:08my host my database is posgress my user
15:12is
15:14this one so you will put user here you
15:17will use the same password that you use
15:19to login into superbase okay so let's
15:22say you have put the password uh port is
15:245432
15:26and that's it and when you set save it
15:29will create a connection into your
15:31postgrace database. I have already
15:34created it so I'm not creating it again.
15:37And once you have created it, you will
15:39see this option you know postgrace
15:40account. So I'm just clicking that and
15:44this is the insert operation and you
15:48want to insert data to your fact orders
15:51table. See we are looking at the first
15:53attachment. Okay. So our first
15:55attachment is this fact aggregate table.
15:58We are ingesting this data into fact
16:02orders aggregate and here you need to do
16:07uh this mapping. Okay. So order ID goes
16:09to order ID. Customer ID goes here.
16:13Order placement date goes here. And here
16:16I want to format this date differently.
16:18So here it is month, year and um
16:22actually date, month and year. But what
16:25I want to do here is I want to say
16:28convert it to date time
16:41when I was doing this two date time
16:44initially I was getting an error. See if
16:46you do
16:47just the
16:49two date time you get error that you
16:52cannot convert to luxen date time. So
16:56the format for the date that we have in
16:59our email. So let's look at that. So
17:02here the format is ddmm and then year
17:05and looks like that luxen I think it's a
17:07javascript library is not able to
17:09understand that. So we need to tell it
17:12okay what format is this? And if you
17:15look at the documentation of this see if
17:19you look at I think two data yeah here
17:24in the bracket you can specify the
17:25format. So they have this documentation
17:28which says you can specify this
17:30particular format where you can say ddmm
17:32y. So I'm going to use this and I'm
17:36going to specify that format. After that
17:39it is parsing that this particular date
17:41into datetime object and then you are
17:44saying format.
17:46So I want to essentially convert this
17:49into this format where you have year
17:52first then month. This is like iOS
17:54standard format. So that way you know uh
17:58the injection into postgrace is
18:00smoother. Okay. Okay. So sometimes you
18:01have to do all this date conversion,
18:03format conversion etc into your this
18:06this mapping that you are creating in N8
18:10and I already map on time info
18:14if and so on. Okay. So test the step.
18:18After testing the step when I go to my
18:21Postgress database I will see data for
18:2417th May. See 17th May. Previously it
18:27wasn't there. I had it till 16. So
18:30through this step it inserted that data
18:35from that file. Okay. So now what I can
18:40do is I can go back. I think this
18:44mapping looks good to me. I will just
18:48copy this because I might need this in
18:50the when I'm doing the mapping for the
18:52other file. Okay. So this one this flow
18:56is good enough. Now we will need to map
19:00the other file. So here this one I will
19:03rename and I will say extract from
19:06extract from file
19:10aggregate. Okay.
19:12And you need to now create another flow
19:16for the second file. Remember we have
19:17two files. So the second file is your
19:21order line items. So here you will
19:26specify your second file here. Okay.
19:30And the test tab. So it is pulling the
19:33data from the second file. See you can
19:35see all these fields. Go back to canvas.
19:39And here add your postgress. Same folks
19:43you are just repeating the same steps
19:46here.
19:48And
19:49here you are specifying your order line
19:52table line item table.
19:56And then just see map these fields.
20:04And for order placement date you will do
20:06the same thing to date time and in this
20:10two date time you will specify
20:15this particular format. Okay. So let me
20:17just copy paste. This is the format
20:21I am using
20:24and then I'm formatting it to this
20:26format ISO format Y
20:29mm DD. Okay. Then customer ID. So make
20:33sure you're mapping
20:36all these fields accurately. I mean
20:40sometimes people might drop one field
20:41into another. So if you make that
20:43mistake, of course it's not going to be
20:45good. So just map it.
20:55As you can see all the fields are
20:57mapped. The only format conversion is
21:00happening with order placement date.
21:01Other fields are okay. And I will click
21:04on test step here.
21:07Input invalid for agreed delivery date.
21:11So looks like we will just use the same
21:15format for these dates because
21:19see it had a date 35 2025. So maybe it
21:23got it wrong over there.
21:25So we are converting it to standard DDMM
21:30Y format first and then converting it to
21:32our standard ISO format. So let's do it
21:35for all the dates. Okay. Do we have any
21:37other dates? So on time is good. Okay.
21:40dates are being taken care of. Okay,
21:43let's test the step now.
21:50I was getting all those errors because
21:52there was some format error with my
21:55Excel file and I have fixed those. Okay.
21:58So, if I pull my files here, CSV files,
22:02then these date formats were previously
22:05inconsistent. So, I was getting it. You
22:06will not get these errors because I have
22:09fixed those issues. Okay. So if you're
22:11using this new files which we are going
22:12to attach down below, you're not going
22:15to get those errors. So now u but I
22:19think it's still a good idea to convert
22:21it to standard
22:23ISO format. Okay. And then do the format
22:26conversation to conversion to whatever
22:29format you need. So we are going to do
22:30it for all the dates here. And once you
22:33are ready, you can taste this particular
22:37step. So let's test this.
22:41All right, it executed it fine. If you
22:44go to your database and if you look at
22:47your order line items, see it is still
22:50refreshing. Right now it's showing 16.
22:52Now it changed. You see it changed. So
22:54it inserted all the records from 17th
22:56May. But I I previously deleted the
23:00records from
23:02aggregate as well. Okay. So I I still
23:04don't have aggregate record because I
23:05was fixing those errors. So you don't
23:07have to run these steps but I'm going to
23:10u just execute this one
23:14so that I see the 17th data in both of
23:17it. Okay. So here
23:20so essentially you want to go to a stage
23:23where you see the data for 17th for both
23:26fact order online and fact orders
23:29aggregate.
23:31Now just in case if you're facing errors
23:33and if you're running it again and again
23:35and if you want to delete the data you
23:36can use this query you can you know if
23:39you want to delete the data let's say
23:41there is some problem you can use this
23:43query you can say delete from fact order
23:45line where order placement date greater
23:48than this so that way you are deleting
23:50that 17th May data uh you you need to do
23:54it for both the tables okay so one for
23:56fact order online and the other one for
23:59aggregate table. Uh so so let me do it
24:02so that you know. So let me delete these
24:05records. Okay. So delete all the records
24:07for
24:09aggregate table first. So these records
24:11are deleted and then delete it for
24:15the um
24:19other table. Which other table we have?
24:22Order online. Right?
24:24So see I deleted these records. And when
24:27I go to postgrace
24:30uh the table view
24:32so it is refreshing you see down below
24:34it is still refreshing but when it is
24:37refreshed you see data only till 16 same
24:40here you need to have sorting by the way
24:42okay don't forget that that's uh see
24:44it's still refreshing so let me just
24:46yeah 16 see and it is sorted based on
24:50descending order so to ingest the data
24:53now I will run the entire workflow so
24:56after deleting it. When I go here and
24:59when I say test workflow, it will taste
25:02both the workflows. See first that first
25:05file, second one is second file. And
25:07when I come here now and refresh, I
25:12should see data for 17th. See 17th May.
25:16Same for the second table. When I
25:18refresh, I will see data for 17th May.
25:22All right. Our data injection part is
25:24over. Now one additional thing I will do
25:26is also ingest the file for USA.
25:30Remember so far we ingested data only
25:32for India. You are given all these
25:35files. So you will find two files for
25:37USA. And I'm going to compose an email.
25:41I'll just send it to myself and just say
25:44daily
25:46sales USA
25:5117th May. And I will just attach these
25:55files. So attach both the files for USA
26:00and send it. So when you send it, see I
26:02see the email here.
26:04But my workflow has not triggered yet
26:07because
26:09it is inactive. You see the workflow. I
26:12gave it a name postgris data injection.
26:14It is still inactive. If it was active,
26:17it will be monitoring my email and it
26:20will be sending it. But you can kick it
26:23off manually as well. So I'll click on
26:27test workflow
26:29and see here it inserted 57 items and
26:34here 109 items.
26:37So let's verify if that is correct. So
26:40USA
26:42aggregate data see 57 item because there
26:44is a header. So 57 items total. So that
26:47is correct. And then order line for USA
26:51is
26:54109. There is a header. So 109 rows in
26:57total. 109. So it inserted that data.
27:03And here
27:07I think here even if you refresh it will
27:09still show 17. So you have to look into
27:11the order ID and stuff like that. But
27:14let's see order line as well. So both of
27:17these tables are updated with USA data.
27:23Let's now analyze this data in
27:25quadratic. Quadratichq.com is a website
27:29you will go here. You can chat with your
27:32data and get insights all using AI.
27:36Click on open quadratic. So you will
27:38come here. You're going to get some free
27:40credits by the way. So don't worry,
27:42you'll have enough credits to run this
27:44project. And then here you will click on
27:48new file. Then let's connect this to
27:51posgress. So here uh just click on
27:55posgress. Okay. And connection name you
27:58can say anything right like atlick m and
28:02then provide that same host name.
28:05You see like when we were here in the
28:08session pooler you need to provide this
28:10host port database user or whatever
28:14right? All these credentials you can
28:16provide here. Password will be the
28:18password for your superbase
28:20and once all of that is done you will
28:23see this kind of entry postgress entry.
28:26So you click on it and you will see this
28:29connection. See it is pulling all these
28:30tables here dim customers etc. You can
28:33also write queries by the way. You can
28:35write select queries to fetch this data.
28:39So let's pull all this data for dim and
28:42fact tables in our quadratic
28:44spreadsheet. So the first one is going
28:47to be dim customers. So let me name this
28:50sheet dim customers
28:53and then you can click on this dim
28:55customer here query selected table and
28:58see it will pull all that data. Isn't
29:01this cool? So we have only 37 customers
29:05and it is running this query with a
29:07limit. Ideally, you should run this
29:09query without limit. But in our case,
29:11it's okay because the number of rows are
29:1337. They're less than 100 anyway. Then
29:16you will create the second one which is
29:21dim
29:22products and check dim products
29:27query select table. Okay. And run it.
29:31Right now I think there is some kind of
29:33issue due to which I'm not able to do
29:35it. So what I usually do is and we have
29:38passed this feedback to quadratic team
29:40they will fix it okay it's just a
29:42temporary thing but just in case if
29:44you're facing this problem click on this
29:46database icon once again click here then
29:50come here then click on query and it
29:54will pull the data. Third one is dim
29:57target
29:59orders
30:01and once again click here click on this
30:05query selected table now here I have
30:09only 20 records so I'm good then fact
30:15order online back.
30:32Okay, in dim target orders, I think I
30:34made a mistake. I'm just saying this
30:37one. Actually, I should be pulling data
30:38from a different table. So, let me just
30:41delete this and
30:45dim target orders. Right. So dim target
30:47order should be this
30:49query. Okay, query is not working. So
30:52let me just go here
30:54again. Say query
30:57and here only 37. So I'm good. The next
31:01one is fact order
31:06fact order line. Okay, fact order lines
31:09which is this table.
31:12And for that particular table once again
31:16connect to posgress
31:19and select that fact order line and
31:22query. Now when I query I will pull only
31:24100 records but there are more actually.
31:27So what you should do is uh remove this
31:29limit clause and execute the query
31:32without limit so that you can pull all
31:36the orders. See how many orders do I
31:39have? See you have so many orders. Okay,
31:41it's definitely more than 100. So you're
31:43going to pull that and same way setup
31:45fact orders aggregate.
31:56See, so I got these many records. So
31:58make sure you are running your queries
32:01without the limit clause. And I realized
32:04that even dim customer table was also
32:06mapped incorrectly. Maybe I selected the
32:09wrong one. So for dim customer make sure
32:11you are selecting the right table. Okay
32:13that's very important. Okay. So see you
32:16are in dim customer uh sheet dim
32:19customer query and that's it. Right. You
32:21can move the table here. So we are all
32:24set. Whenever you are doing this kind of
32:25analysis you almost always need a date
32:28table. A table which has the required
32:30dates with extra columns for months year
32:34etc. If you have worked as a data
32:35analyst you will know the importance of
32:37this dim date table. Okay. So I'm going
32:40to create
32:41this dim date table. And the good news
32:44is that with AI you can just type a
32:47prompt here and create that table with
32:49all the required prompt.
32:56Okay. So that it's very easy. So I'm
32:58going to copy paste this prompt for
33:00creating the date table. and make sure
33:04when you're running this prompt you have
33:07this as a active sheet. Let's say if you
33:10select this then you see it will insert
33:12a table in that sheet. You don't want
33:14that. Okay. So click here. So that way
33:18you see dim date that is your active
33:19sheet where you're working. This is your
33:22uh cursor location and you are saying
33:24that create a date table that has dates
33:27from this March 1st to March 31st.
33:32And when you do this, it will use AI to
33:36first write Python code and it will then
33:39execute that Python code. And you can
33:42see the output. See date table. It's so
33:44awesome. So you have dates from 1st
33:48March to whatever. And you have separate
33:51columns for year, month, day. All of
33:54these are useful when we will do the
33:56analysis later on. You can also see the
33:58code. So if you click on this icon you
34:01can see the code and if you know Python
34:04coding if you want to let's say change
34:07certain things you can modify the code
34:09manually or you can chat here that okay
34:12do this particular change at this line
34:14or change the format of this column and
34:16so on. The next step for our analysis
34:19will be to create the exchange rates
34:23sheet. Okay. So I'm going to create
34:27exchange
34:29rates sheet and this sheet will uh
34:33contain the exchange rate conversion
34:36between USD and INR. Okay. And that
34:38conversion is required because later on
34:41we need uh this for performing our
34:44analysis. See if you want to say okay
34:46what are my sales number in USD and what
34:49are my sales number in INR then you need
34:52some kind of conversion right because
34:54our business is in both the location
34:56India and US and you notice that we have
34:58sales number coming in for both the
35:01countries in our product table also see
35:04we have INR and USD price so you know
35:08this conversion the exchange rate table
35:11that we have it can be useful now how do
35:13you get the exchange rates. Well, we are
35:17going to use this website called
35:19openexchange rates.org to get the actual
35:22conversion rate. Okay, we are not going
35:24to use some dummy one. So, you create an
35:27account using your Gmail or whatever
35:30whatever way you want to create account
35:31up to you. And when you go to your
35:33dashboard, you will create this app ID.
35:37Okay? So you will create this app ID
35:40which is sort of like an API key and you
35:43will use this in that quadratic. Okay.
35:45So make sure you have the app ID copied
35:48somewhere. So you copy it and then you
35:51are going to use a prompt. So once again
35:54the prompt that we have given you. This
35:56is the prompt. See create an exchange
36:00rate table from this date to that date
36:03and use this open exchange rate API.
36:07Okay. So you're telling that your LLM to
36:10kind of create a code for it. So let me
36:13just copy paste here.
36:16Copy paste here and make sure once again
36:19exchange rate is the active sheet. I
36:22will supply here. By the way this app ID
36:24you see this app ID here in the prompt.
36:27You will use your own app ID. Okay. This
36:30app ID we are going to delete. So it
36:31will not be valid. So it will not work.
36:33So replace this app id with your app id
36:36that you have created on open exchange
36:39rates.org
36:41and after that you will write uh you
36:45will give this prompt it will generate a
36:47python code. So let's look at the python
36:49code here.
36:51You see this is the python code that it
36:54has written and it is executing it right
36:56now. All right. How cool is this? I see
36:59the exchange rate table. This is USD to
37:03INR rate. Okay. And if you want to make
37:06any changes in Python code, feel free. I
37:09know Python coding. So I can make a
37:12change such as the exchange rate. I need
37:14it only till four decimal precision.
37:18Okay. So I can change the code manually.
37:21But let's say if you don't know Python
37:23coding, you can ask here that I want the
37:30exchange
37:32rate
37:34to be in four decimal precision and it
37:39will update that code and it will rerun
37:42it again. I still believe having coding
37:45skills is useful because sometimes you
37:48get give this prompt and let's say see
37:51it made some code changes. Okay, see it
37:53is rounding it to four decimal. I'm
37:55accepting but I know coding that's why I
37:58can say okay it's a correct change. If
37:59you don't know coding at all sometimes
38:01you know you might get bad result. So I
38:04believe having coding skills can still
38:07help you although we are doing automated
38:09coding through uh this AI but anyways
38:13this will rerun it again and you will
38:15see the new result. Okay you see these
38:18numbers are in now four decimal
38:20precision. So exchange rate table is
38:23created. It's looking good. Now let's
38:25move on to the next step which is doing
38:28data cleaning and summarizing required
38:30data in one table. We are going to
38:32provide you this document. So you'll be
38:34able to copy paste this entire prompt
38:37which is marked in yellow. So let me
38:39copy paste this entire prompt here
38:43and we will run this prompt in a new
38:46sheet. We'll call it fact summary. So in
38:50fact summary
38:52let's run this prompt. Okay. And when
38:55you run this prompt uh after some time
38:58it will create uh this particular merge
39:01table. Now what exactly was this prompt?
39:05So if you have done data analysis
39:07you know that you need to merge all this
39:10dim and fact table into one uh
39:13denormalized table which contains all
39:15the columns. Okay. So we are saying that
39:18load data from fact table this dim table
39:21exchange rate table and then clean the
39:24data. Okay. So you're cleaning uh
39:27product see converting product ID and
39:29customer ID to numeric because they were
39:31string then uh removing wide spaces
39:35because usually in the real life data
39:37sets you will find all these white
39:38spaces null ids you are converting ids
39:42to integers dates to date time. So if
39:44you have worked in data analysis, these
39:46are the typical steps that you follow
39:50okay like cleaning data then merging the
39:53table and then you will say okay merge
39:55these tables using this column okay
39:57product ID column customer ID column and
40:00so on. Then you will also calculate the
40:03total amount. See you are doing the
40:04currency remember we did that currency
40:07conversion USD to INR rate whatever. So
40:10you are doing that uh conversion and
40:13then you are producing this uh final
40:15output and you will see this kind of
40:17table at the end. All right, you can do
40:19some uh testing. You can look at some
40:21couple of rows and make sure your data
40:23is correct. But the way I'm seeing it
40:27right now is we have one merged table.
40:30Not only merged table but we have done
40:32cleaning as well. And in the code that
40:35was generated um if you look at it see
40:38it is using data frame and it is doing
40:41some internal cleaning.
40:44You see if you know pandas and python
40:47see it is doing conversion to date time
40:50then it is merging it is creating some
40:53calculated columns and finally it is
40:55creating this uh final fact summary
40:58sheet.
41:01All right, we will now begin our data
41:03analysis session. Here with me, I have
41:06Hamand Vial who has worked as a data
41:08analytics manager in Europe for more
41:11than 7 years working on multiple large
41:14scale data analytics projects. He's also
41:17a supply chain domain experts. So the
41:20goal here is to teach you not only data
41:23analytics using AI but also teach you
41:25important supply chain concepts. Over to
41:28you Haman. Thank you D. That's very
41:30exciting to be here to work on a supply
41:32chain project and I have seen the data
41:34that we have created from quadratic and
41:36we have pulled the data from superbase
41:38postgrade and the next step is to create
41:41supply chain KPIs. Correct?
41:43Yes.
41:45So before we create the KPIs as a data
41:48analyst the first thing to do is to
41:50validate the data that you got on this
41:52quadratic spreadsheet with what we got
41:54in the superbase.
41:56So let's do that step first. So I'm
41:58opening this sheet. Okay, I'm going to
42:00this dim customers table. So here we
42:03have 37
42:06which means 35 rows because the first
42:08two rows are headers. And let me go to
42:10the superbase postgrade database.
42:14So here we have 35 that's exactly
42:16matching. So we are good. And then we do
42:20the same thing for products. We have 18
42:22here. We have 18 here as well. That's
42:24good. And then we have targets. Target
42:26orders. We have 35 here and 35 here as
42:29well. That's also good. So now let's go
42:31to the fact order line. So in here the
42:34sheet we have around 23 554 and we have
42:3925,538.
42:40Okay, that's not matching. So with
42:42respect to orders aggregate we have
42:4513652 here in the database that we have
42:49got we have 13314 that's not matching as
42:52well. And if you see further, okay, it's
42:57adding this 13652 rows. It's it's it's
43:00present there. It's simply
43:03showing it as null. So I investigated
43:05further with this. It seems there is
43:07some, you know, intermittent delay in
43:09fetching the data. I have reported this
43:11to the coordinating team. They are
43:12working on it. But I think we should
43:14consider this as actual data. Whatever
43:17we see in this table as actual data and
43:19proceed with our analysis. But folks
43:21when you're watching this please know
43:23that there is this gap
43:24and the idea of this video is to teach
43:27you this evolving
43:29data analytics tools through AI. So you
43:33want to understand this uh AI mindset of
43:37doing data analytics. So even if the
43:40there is some issue with the data it's
43:42okay these are intermittent issues which
43:44will get fixed. The purpose is to learn
43:47and evolve along with these tools. All
43:50right, let's begin looking into the
43:52KPIs.
43:55Okay, these are the seven KPIs we are
43:57going to create. We could see this on
43:59the screen and I tried something very
44:02interesting with quadratic. I just
44:04simply pasted this KPIs there in the
44:06sheet without even explaining how to
44:08calculate it. So if it has to calculate
44:11it correctly, it has to go to the
44:12internet, understand this metrics and
44:14then create the results. Let's see if
44:16that works.
44:17I've copied this prompt.
44:20So all these prompts will be given to
44:22people so they can replicate the same.
44:24And we went to the quadratic sheet.
44:30Then I just simply pasted this prompt.
44:34So I'm going to create a new sheet for
44:36this. I'm going to call it like APIs.
44:43Yeah. Let's see if that works.
44:48It's very interesting, right? It fetched
44:50all these results. But how do you know
44:52if this is correct? You need to know two
44:54things for this. First, you need to know
44:55the data analysis because you can open
44:57this Python code and check really if
45:00this thing is
45:02if this code is making sense or not.
45:04Even if this code is making sense, you
45:06need to have some domain knowledge to
45:08understand if the calculation is right
45:10or not.
45:10Yeah. So, can you explain what these
45:12KPIs are? And can you increase the font
45:14size so that I can see better and then
45:16you explain what exactly is the meaning
45:19of these KPIs?
45:20I'm going to explain this with a very
45:21simple example so that everyone can
45:23understand and having this kind of
45:25domain knowledge is very important you
45:27know especially if you're targeting your
45:28career in operations and supply chain.
45:30So I've created this very simple table.
45:32Just imagine you made an order in
45:34Amazon, right? You made it on 19th of
45:37May and you made the first order. So it
45:39is called as order number one. And you
45:41ordered keyboards and you ordered five
45:43keyboards, right? The next day you made
45:47an order but this time you ordered two
45:49items which are keyboards and mouse. So
45:52now tell me based on this table
45:56how many orders are there?
45:58There are two orders right? want order
46:00one and two.
46:01Yes, there are only two orders. On first
46:03day you made only one order. Second day
46:05also you made only one order. So there
46:07are two orders, right? So that's two.
46:10And how many lines are there?
46:12Three if you think about it. Three.
46:15Yes. So that's exactly the difference
46:18between orders and order lines. Order
46:20lines will always be more. Orders are
46:22nothing but whenever you place an order,
46:24it it will get an order ID and that's an
46:26order. In each of this order you might
46:28be placing multiple items. So those will
46:31create order lines. Okay. So now let's
46:34understand what is line fill rate and
46:36volume fill rate. Okay. So we clearly
46:39understood what is order and order
46:40lines. Now with the same table let's
46:43just expand it. Right? This is the
46:45quantity order and this is the quantity
46:47delivered.
46:49So now tell me is this order delivered
46:51in full or not?
46:53Yes, because I ordered five and it was
46:55delivered five.
46:57So instead of writing yes, let's do
46:58binary. Let's write one. If the order is
47:00delivered in full, let's write one. If
47:02not, let's write zero. Okay. And tell me
47:06for the second line,
47:07no.
47:07This is in full or not? No.
47:09No,
47:09it will be zero. And for the third line,
47:12one.
47:13This is exactly how it is written in a
47:14typical supply chain table as well.
47:16Okay.
47:18So we have total three lines, right?
47:20Line one, line two, line three. Out of
47:23these three lines, how many lines we
47:25delivered successfully?
47:27Two.
47:28Two. Okay. Because one and two, there
47:31are two. And uh so what will be the line
47:35fill rate now? Can you make a guess?
47:36It is 2x3.
47:38Exactly. So this is line fill rate
47:40percentage. It is 2x3 because
47:44you have to find a ratio of total lines
47:48you've delivered which is two divided by
47:52total lines order
47:54which is three. So your line fill rate
47:56is 66.67%age.
47:59So now when it comes to volume fill rate
48:01let me make it.
48:07Can you make a guess what could be a
48:09volume fill rate?
48:12M
48:15for volume you have to take into account
48:17the quantity
48:19exactly. So here let's take a sum.
48:23So what is the sum of quantity ordered?
48:26Right? It's 17. And what is the sum of
48:29quantity delivered? It's 17.
48:31So the volume fill rate is nothing but
48:33the ratio of quantity delivered to the
48:35quantity ordered. You just take this
48:38number and divide it by this number.
48:43So it's 85%age.
48:45So that's the volume fill rate. And in
48:47supply chain line fill rate and volume
48:49fill rate is a very important metric for
48:52supply planners, supply planners and uh
48:56supply managers and production managers
48:59because for them it's very important to
49:01know how many lines were ordered and how
49:04much they managed to deliver. So this is
49:06how the performance is evaluated.
49:09And this volume fill rate is also very
49:11important for sales people. It's
49:13important for supply people but also
49:15very important for sales people. When
49:16they're having a negotiation or some
49:18kind of discussion with the customers,
49:20they will say, "Hey, last year you
49:22ordered uh 10 million quantities. We
49:25delivered 9.98 million. So we we made a
49:28you know we almost fulfilled all your
49:30orders." because this conversation is
49:32directly proportionate to the best deal
49:34they can get from the customers. So this
49:36metric is very important. Okay, let's
49:38move on to two other important metrics
49:39which is on time and info.
49:44If you expand this table to two more
49:45columns, two more information which is
49:47agreed delivery date and actual delivery
49:49date. So can you now tell me if this
49:52order is delivered on time?
49:54Yes, because the both the dates are
49:55matching.
49:56Perfect. So it's one and for the second
49:59row it's one as well. for the third R1
50:01as well. Okay. So, how many orders were
50:04delivered in full?
50:05Delivered in full two. We we have
50:07discussed that, right? So, two orders.
50:08Actually, it is not two orders because
50:10if you think about orders, this is where
50:12most people make mistake.
50:14They think about lines here. In full is
50:16calculated at order level, not at the
50:18line level.
50:19Oh, only one order because order one was
50:22delivered full but order two was not
50:24delivered full. Was not delivered. Good
50:26point. In terms of on-time orders,
50:30on-time orders were two. Both the orders
50:33were delivered on time. Okay.
50:36So this on-time orders it is used by can
50:38you guess who uses it in the supply
50:40chain field? So there are different
50:42supply chain departments.
50:44No, I I don't know.
50:46Okay. So it is used by warehouse and
50:48distribution people. So these are the
50:50people who are in charge for delivering
50:53the order on time. So you know all the
50:56shipments all those trucks going on. So
50:58this is managed by the warehouse and
51:00distribution people. For them this
51:01metric is very important. It's called
51:03warehouse and distribution. And this
51:06info is again used by supply
51:09managers more at regional level. So they
51:12want to see this number at a regional
51:13level. Okay.
51:15Mhm.
51:16So now we are going to calculate the
51:17most important metric which is on time
51:19in full percentage. In supply chain
51:21terms it is called as. What if is
51:24something very very commonly used metric
51:26and for this you can see the data is
51:30consolidated at the order level not at
51:32the item level. So you see we have three
51:34lines here but we here we have only two
51:36lines because it's at the order level
51:38and you can see the values are also
51:39consolidated here for the second order
51:41the total quantity is 15 and total
51:43delivered is 12. You can see the same
51:45numbers here 15 and 12. So at this level
51:50this order is not delivered in full but
51:52it is delivered on time. So now tell me
51:54if this order is delivered both on time
51:56and in full.
51:58Yes. It's a end condition between E and
52:00H column.
52:02Yeah. Then it's one. Tell me for this
52:05this one is zero.
52:06Yeah.
52:08So I'll tell you a very interesting
52:09point. For one particular order there
52:11might be 200 lines. A customer might
52:14place 200 items in one particular order.
52:16Let's say they delivered all the 199
52:18items in full but one line they did not
52:21deliver in full. Even they miss one
52:23quantity. That order will miss if
52:26[Music]
52:28it's it's a very harsh metric. It is it
52:30is super super harsh. Sometimes the
52:32order might have like 200 300 lines and
52:34even if they miss one line that order is
52:36failed. But that's how it's it's
52:38calculated in supply chain and uh so
52:41let's calculate the in full percentage.
52:42So it's very uh you know very easy to
52:44calculate. So now we know only one order
52:47is delivered in full. So it's nothing
52:49but 1 divided by this value two orders.
52:54So it's 50%.
53:01And here total on-time orders are two
53:04and the total orders are two.
53:07So it's 100%. This is super rare in in
53:10in the real supply chain world, but for
53:12this example, yes, this is possible. and
53:15on time and inflow. So this is nothing
53:18but how many orders we got both on time
53:20and info only one right out of two
53:23orders
53:28it's also 50%.
53:37So this on time and in full percentage
53:39is normally used by supply chain
53:41directors or supply chain VPs. So this
53:44is the metric that they will focus on.
53:46They will not focus on any other metric.
53:48So this on-time and infill percentage is
53:51also called as reliability.
53:54Reliability is also a very important you
53:56know term in supply chain. So if
53:58somebody's asking what is the
53:59reliability? If you say 85%age which
54:01means 85%age of the times you will
54:04deliver the order in full and on time.
54:07So the customer will ask you what what
54:10is the reliability I can have? If you
54:12say 90%age that is really good which
54:14means out of thousand orders they place
54:16you are sure to make that 90 90% of the
54:20times the orders will be always
54:21delivered on time and in full.
54:24So based on this thing the service level
54:26agreements are made and the contracts
54:27are dealt and even the you know there
54:30are a lot of commercials involved in it.
54:32All these things are based on this. So
54:34this is also the point where the supply
54:36chain folks the sales and marketing
54:38folks collaborate. They want this number
54:41to be really really high. Only if this
54:42number is high, there is a very high
54:44possibility of customer retention. There
54:45is a very high possibility of getting
54:47better deals and all this stuff. I
54:50always wonder when I order things from
54:52Amazon like how how well they take care
54:55of this entire supply chain because they
54:58order things from China and here in my
55:00home in US when I order things get
55:03delivered on a very next day and by
55:06learning all these domain concepts now
55:08I'm kind of getting better understanding
55:10of their inner world
55:12on you know companies like Amazon would
55:14have very strict
55:16uh uh strict rules on on metrics like
55:20Otif
55:22and their suppliers will have to have
55:24very high like you know high performance
55:27when it comes to percentage etc. So I'm
55:30really glad that I'm learning this
55:32domain concepts because in the world of
55:34AI where technical things are being
55:37automated the knowledge of these domain
55:40con concepts is something that can set
55:42you apart from the competition when it
55:44comes to job market and career growth.
55:47Yes, Amazon is a good example. So,
55:48reliability is one important metric
55:50where you always promise to deliver like
55:52a perfect order with which is on time in
55:55full with perfect documentation. But
55:57there is also another metric which is
55:58like the the response cycle time like
56:01when you place the order or the order
56:02fulfillment time. So that is something
56:05which Amazon worked on. Earlier it used
56:06to be like three or four days in in the
56:08e-commerce world. Amazon made it like
56:10same day, one day and now you see hyperd
56:13deliveries which is happening in like 15
56:14minutes, 10 minutes and all those
56:16things. All this is based on this metric
56:18and uh so what you spoke about is an
56:20example of B2C where a business is
56:22giving directly to the consumer but this
56:25this particular concept is mostly tied
56:27to B2B where there is one company which
56:30is having a lot of distributors and who
56:32are having their consumers which makes
56:34the supply chain even more difficult
56:35because they also have their supply
56:37chain. they have to maintain their uh
56:39you know they have to maintain the
56:40promise to their consumers and it it
56:42gets very long and most of the times
56:45this company is also procuring from some
56:47places like China and you know this is
56:49why right like uh if you are getting a
56:51product in your hand this mouse if
56:53you're getting it today it it is
56:55probably planned like some 300 or 400
56:58days before before it comes to your hand
57:00so that's that's very interesting that's
57:01why I love supply chain it's is a very
57:03interesting field
57:04many of the people who are watching this
57:06video are the uh customers of uh you
57:10know things like Blinket, Zomemetto,
57:14even Flipkart and all these companies
57:17even though it's B2C internally they
57:19also work with their their business
57:22partners. So now you all are getting
57:25some insights into that beautiful world
57:28of supply chain.
57:30Great. But supply chain is super vast.
57:32If uh if folks you're watching this
57:33video, if you're more interested, try to
57:36uh you know type SC score on in Google
57:40and uh search for this. It's supply
57:43chain operational reference model. It
57:45has a lot of interesting concepts. You
57:46can you will definitely like it. Okay.
57:49Now let's get back to this quadratic
57:51sheet and check the formulas. I would be
57:54really surprised if it has calculated
57:56all these formulas correctly because we
57:58gave no reference. Right now you know
58:01the supply chain concepts. You can also
58:02validate this along with me. Let's first
58:05check the order lines. So it says
58:09calculated from the fact order line.
58:13So yeah total order lines is nothing but
58:15the length of order lines. What is order
58:18lines? It has calculated it from the
58:21order lines. It has went to this table
58:25and it has calculated this which is
58:27correct. This order line table contains
58:29all the orders. So it did not take the
58:31unique value of dollars. So it knows
58:33unique value of orders means total
58:35orders. Order lines means it has to take
58:37the entire order lines. So it is
58:39correct. And let's see total orders. So
58:42it says total orders is nothing but
58:44length of order aggregate. What is order
58:47aggregate? Okay. Oh, that's really good.
58:50It somehow understood this fact
58:52aggregate table contains a consolidated
58:54order. If you are wondering just just
58:56remember this example right these are
58:58order lines then I consolidated it right
59:01two orders this is the same example here
59:03here we have all the order lines and
59:05this aggregate is nothing but
59:06consolidated orders so it it it did a
59:08really good job there I'm I'm quite
59:10surprised and
59:13so let's see the line fill rate so again
59:16it's using the order lines table and in
59:18this it is calculating where the values
59:21in full is equal to one which means the
59:23orders which are delivered in full which
59:26are delivered in full quantity and it is
59:29dot mean it means it's it's it's a ratio
59:33it's it's basically dividing that value
59:36with the rest of the values which is
59:37correct. So it it's taking a ratio of
59:39values that is one versus the rest of
59:41the values which is zero. So this
59:44calculation is also right and the same
59:47thing goes with volume fill rate that is
59:50that is also correct. So it calculated
59:52the total delivery quantity with the
59:54order quantity sum. That's perfect. And
59:59with the on-time delivery yes it has to
1:00:01see on time and in full and all these
1:00:04three things has to be calculated at
1:00:06order level. So you see it uses order
1:00:09aggregate table which is nothing but the
1:00:12consolidated order level. I'm I'm really
1:00:14surprised that this is able to calculate
1:00:16it so accurately without us giving an
1:00:18external references. This is this is
1:00:20good. If you're not using AI tool, you
1:00:23would have spent few hours and now you
1:00:27you did all this work in few minutes. So
1:00:30you can realize the productivity gain
1:00:32here.
1:00:33Yeah. But this also reinstates the point
1:00:35that you cannot miss the fundamentals,
1:00:37right? Let's say if you don't know the
1:00:38supply chain concepts, if you don't know
1:00:39the basics of Python, you won't be able
1:00:41to validate this code. You would have
1:00:43given this to another AI to validate,
1:00:44but you don't know even if that is
1:00:45correct. And this will go on. So you
1:00:48need to know the basics and you can kind
1:00:50of you know maximize your productivity.
1:00:52That's how I see using the ZI tools.
1:00:54AI models will hallucinate on occasions
1:00:57and if you are you know deploying this
1:00:59code to production or using it to make
1:01:02important decisions. It is important
1:01:06that you know the fundamentals. So you
1:01:08all might have a question of should we
1:01:10learn Python? Should we know uh data
1:01:13analytics? Should we know domain
1:01:14concepts? Yes, of course you need to
1:01:16know because you are the one who will
1:01:19validate if AI generated correct output
1:01:22or not and it will make mistakes folks.
1:01:24See right now it did a good job but
1:01:28let's say one out of 100 times it is
1:01:30making a mistake then that mistake can
1:01:32cost your company a big money. the
1:01:34company will hire you because you know
1:01:36all these fundamentals and you can work
1:01:38with AI and you can figure out its its
1:01:40mistakes and you can fix it and you can
1:01:42also guide AI you know typing that
1:01:44prompt and you can guide AI to do the
1:01:48right thing
1:01:50okay now we have seen how to create
1:01:52these KPIs right so next why don't we
1:01:55ask some business questions and see if
1:01:56it is able to answer it properly
1:01:59one thing I'm curious about is if there
1:02:01is a way to track monthly on-time
1:02:03performance months.
1:02:04I mean, yes, I think it should be able
1:02:06to do, but let's see how it is
1:02:07responding, right? I'm just simply going
1:02:09to ask uh show me monthly
1:02:14on time performance by cities.
1:02:18Mhm.
1:02:20Cuz uh if you're a VOS and distribution
1:02:22manager, you would like to know it by
1:02:23cities.
1:02:25And let me do it in a new sheet.
1:02:28I'm going to call it
1:02:30business questions.
1:02:35you press enter.
1:02:39Okay, it generated a chart. I think
1:02:41that's pretty good from this. I can
1:02:42easily see how the trend is declining
1:02:45and uh it's showing me pretty good like
1:02:47how I would expect to get created in a
1:02:50PowerBI or something. That's that's
1:02:52good. So maybe I can ask more questions
1:02:55to it like uh I've created a prompt. Let
1:02:58me open it. So I want to ask like show
1:03:00me the top five customers based on their
1:03:02order value and their on-time percentage
1:03:04in full percentage on time in full
1:03:06percentage. I want to say also add the
1:03:08customer name, customer ID and city in
1:03:10the table. So let's see what it
1:03:12provides.
1:03:13I'm going to copy this
1:03:15and folks you can mention uh in a prompt
1:03:18if you want chart or a table.
1:03:21you. Can I just read this one more time?
1:03:31Okay, I pasted the prompt here in the
1:03:33chat window and uh let's see what
1:03:36happens.
1:03:39And wherever you have active focus in
1:03:41the sheets, see you have it at row
1:03:43number 26. So I think that is the
1:03:45location where it will insert the the
1:03:48new visual.
1:03:50Yes. Okay, you can see it generated the
1:03:52data here and uh it did not generate in
1:03:55the line 826. It generated you know
1:03:58somewhere randomly wherever
1:04:01and uh that's good. It just gave me
1:04:04everything what I wanted. It gave me the
1:04:05customer ID. It gave me the customer
1:04:08name, city, total order value on time
1:04:13info percentage in full percentage and
1:04:15on time percentage. Let me quickly check
1:04:19the code. Okay, it's taking the fact
1:04:21summary for this which is good. So if I
1:04:23quickly skim through this code, it looks
1:04:25good. But if I have to do this for real,
1:04:27I would I would spend some time with
1:04:28this. I would spend another 10 to 15
1:04:30minutes in this to check. But at the top
1:04:33level, it looks correct. Yeah, I think
1:04:35uh that's pretty much good. But as for
1:04:38you know like uh here I ask for top five
1:04:40customers. It seems all my top five
1:04:42customers are in US. So let me copy this
1:04:45same prompt here and see if it is
1:04:48consistent. I'll put it here and say
1:04:51show me top five customers in India.
1:04:53I'll just add one more line here
1:04:56and see what result it provides.
1:05:02I think that's very consistent. It gave
1:05:04pretty much the same results but with
1:05:07Indian customers. That's good. That's a
1:05:08good summary table.
1:05:10Yep. Looks amazing. You just type a
1:05:13prompt and it is doing most of the work
1:05:15for you.
1:05:16Yeah, I I I mean it's it's amazing. But
1:05:18at the same time, I would say this also
1:05:21reinstates the point to me. The
1:05:22fundamentals are very important. You
1:05:24cannot simply work with a tool like this
1:05:26without knowing the fundamentals of
1:05:27Python coding or knowing about supply
1:05:30chain. Not having any domain expertise,
1:05:32it's not going to help. That's why we
1:05:34always stress that having domain
1:05:35knowledge and having the fundamental
1:05:37understanding of this coding really
1:05:39helps a lot. Okay. I highly recommend
1:05:41you to practice this along because
1:05:42quadratic is free. They provide some
1:05:45free AI credits. There is no excuse for
1:05:47you. You need to practice this. And they
1:05:51also promised the quadratic team also
1:05:53promised to provide some special
1:05:54discount for the learners of this
1:05:56channel. We are putting that in the
1:05:57description. You can use that and get a
1:05:59special discount if you are taking a pro
1:06:01account. So like I said, please do
1:06:03practice and also know that this is not
1:06:06the end, right? There are a lot more to
1:06:07explore, a lot more to learn and we are
1:06:09planning to bring more videos of this
1:06:11kind where we are just going to provide
1:06:13a very you know unbiased and neutral
1:06:15view of the tools because they are in
1:06:17the evolving space. We just want to
1:06:19break down the hype and show you the
1:06:20reality and I hope you found this video
1:06:23really useful.
1:06:24This is the end and I want to mention
1:06:25very important thing which is exercise.
1:06:28In the video description below when you
1:06:30click to download files you will find a
1:06:32file called exercise.pdf.
1:06:35Please look into it and work on that
1:06:37exercise. If you have any questions,
1:06:39there is a comment box below. If you
1:06:41like this video, please give it a thumbs
1:06:42up and share it with your friends who
1:06:44are learning analytics and AI.
1:06:47[Music]

About this transcript

This page contains the full transcript of Data Analysis using AI Tools | Quadratic | N8N | Supply Chain by codebasics, generated from the public captions YouTube serves with the video. The transcript has 10,289 words across 1,502 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.