 All right, so can you guys hear me cool? All right Okay, so Let me just start by asking how many of you guys here are business analysts Like you deal with Excel you have a lot of visualization to do, you know People come to you with like Excel files. Hey, can you get something done? And then you maybe at the same time you have to also Maybe at the same time you also use Jupiter any one of you are at their early stages of Learning map plot lip learning seaborn about visualizations cool, cool, cool. Okay, so My stance is just very basic When you run like different kinds of analysis for your data, right? Especially if you want to join Kaggle competitions, and then you try to show people your data most of the time you have your charts built up in Jupiter notebooks and You kind of like export the The images right and then you try to show to be different people online. So I'm here to suggest a way For you to overcome that so that you can have more interaction People can play with your data so that they can know your analysis better. All right, so because I Think as a fellow Python user I understand that you know a lot of your hard work goes into making one visualization, but People will come on come in and ask questions like hey, so can you give me this? Can you give me that and you need to tweak stuff? Right? So it takes a lot of effort. It can't do it at the same time You know, I don't want your effort to go to waste so That's just me Can't see me that very well, but here I am So I am the analytics lead for a skincare company called Paula's Choice Singapore We're primarily e-commerce based and so I run like basic. I run dashboards. I have conducted like Excel and Python workshops before and basically my aim is to implement business intelligence as well as running data science within the organization So I'm also an instructor in port education. So we have Upcoming like also power BI Python Excel workshops going on. All right So to tell you story, right? You kind of usually need more than one chart, right? So this is the part where I think if you use Jupiter or you know programming a lot, you kind of lack of it How why what do I mean by that? Right? Let me just give you a quick Demo what I mean. So like, okay sure, you know, Jupiter anyone uses Jupiter no books. Okay Are you sorry? Are you just starting to use just learning my plot live those kinds of stuff? Are you fairly experienced? Okay, cool, so like You know, you kind of try to run some basic analysis and you try to you know get a kind of like Story from the data, but most of the time when you show this to Someone or you put this in a slide and the person will be like, okay, so what right? Oh and the problem is like What if the person comes in and ask can I do this in a yearly basis? Can I see how this changes by year? Can I see how this changes by country? You know, I think If you are just starting out to learn Python that can be a bit overwhelming What are some of the workarounds right so some of the workarounds may actually be things like You know from what I've been researching so far, you can use Altair you can use You can use plot Lee. I think plot Lee to a large extent achieve what I'm trying to do here So if you are not very interested you can kind of like learn about that as well But you know, even if you were to learn a new package, right? Number one the effort is so much Number two the interactivity is like sometimes, you know limited to this. Okay, sure You know, what if you have other charts involved, right? What if you have other kind of data sets involved? So that's pretty much where I'm coming from and Here I Think most of us can probably or at least people like business analyst, right? Will most of the time think hey, I can just do this on Excel anyway, right? I can just pop the table pop the chart and try to just link everything out I don't need to learn a new package to do this. So How do I So what is my suggestion my suggestion is to park these things that you work on Python into power BI Because if you have used Excel before you are going to be pretty much familiar with power BI You can just drop the visualizations right away. So what do I mean by that, right? Let me just give you a quick example so Okay Okay, so if you guys recall this is the Perplot that we ran earlier and This is also like the correlation graph Perform I mean shown in a heat map and these are the other things that you typically as a business analyst would need to use right, you just get like some kind of Excel file and then you have all these data and then you know, usually you just show a Sales by year that kind of visualization so Most so what you can do here is if you build these things on power BI right when you Have like a manager coming in and asking. Hey, so how does this person do? You know for the sales in that particular region the person that I mean the regions that the person is in charge then Basically, I think it's because the person doesn't have any yeah, so basically What you can do is when you zoom into a particular seller If you notice the seaborn plots and also the heat maps they all change according to this specific scenario Which I think is a very huge value at for someone who you know does most of your hard work on seaborn on Jupiter notebooks, so If I want I can also click on you know different segments and I can it takes some time to load but Basically, you can see how things change right and you can and okay So and what is exactly the code involved in this thing for this chart? I don't know it. Can you guys see it? It's that simple import import my plot lip You have your data sets by default created already and then you just SNS pairplot right it's that simple and For beginner beginner Python users, you know, it is just really really very helpful And you you can just kind of like share this file with other people for them to use Any questions so far? Okay, so Let me just feel how does it generate the plot? You mean this one This one. Oh, so this one Yeah, so this is another one Basically all the work that you have on a Jupiter notebook right or all the work that you spend creating this visualization on Well on Python you can pretty much just translate the code Like all the stuff. Let me show you So you kind of like experiment with stuff On Jupiter right and then this is all the code that you have and you end up building this visualization so what you then do is you create like a a Like a container for Python and then like you drop it into Into the blank space Over here you decide what data is required to plot the graph so This is like a sales data. You can you put in discount or all sorts of stuff and then all you need to do which is What you get here? Sales discount profit, you know shipping costs So these are the things that you need to plot right? so from there the code that you need to Chuck into Power BI is this You just pretty much like copy paste from Jupiter notebook and then you just need to Change the name of your data frame from whatever that you called it earlier into The so-called data set so data set equals the things the Python data frame that You put into this table here like really almost zero work involved and The person who needs to see this visualization You know if you're a data scientist or even if you're just a business analyst learning the basics of Python, right? This will come in very handy You can do whatever you want You can just change all the you know the usual line charts and change the bar charts and The person can go in and just interact with it. So This helps a lot with the way you your stakeholders interact with your data and You know, you don't have to come you don't have to have your colleagues come in and ask Hey, you know, can I get this for country level? Can I get this for a monthly basis, right? If I need to I can just do a slider of the different periods and then all the Visualization here will be updated according to what you need to show Yes Yeah, so far I think so far. That's what I do. I don't so this is This feature was just announced by Microsoft in August last year and a lot of it is still being developed but by and large For you know a quick and re-visualization, I think this achieves What is needed? And I think you know for a company if you are running your own business or if you just want to start out like even just Tackling some of the Kaggle competitions by doing visualization, right? You don't need to restrict yourself to just Plotting visualizations on Jupiter. This is a lot more efficient. This is a lot more interactive and I think that's one Way for me to demonstrate Yes Oh, no, so the good thing about this Power BI right is that you do you it relies on the Modules that you have installed on this PC. So you don't need to worry about like Whether or not Python support and something power BI supports Seaborn by and large they work. It's just that you should try you should avoid having very complicated scripts inside my recommendation is to have an output of your machine learning you know algorithms and Put them in here so that you can very quickly run visualizations for the end user to interact. Yeah, that's my use case here I do not Encourage you running like, you know tensorflow or whatever in here because for the time being Power BI will not Continue will stop the calculations that run over five minutes So, yeah, don't expect to do like full ML work over here But for a very quick and dirty visualization, it gets a job done Yes I So for that you kind of need to pay the pro version, but it's just five bucks a month. So It will support the visualization on the the web interface So, let me just see if I have an example up here Okay, yeah, so Yeah, I don't have it here, but if you have the pro version you can show it so But I think, you know, even even if you don't have the pro version If you want something very simple, you know, kind of like spree, right? Okay, so So if you so Because this is mostly for business analytics purposes I would recommend getting the pro license and just, you know, putting it up So So far what I have done is to publish a link for my Colleagues to visit so I mean I myself have not tried Or basically use this in a corporate environment. So I don't know how that actually works out, but I Haven't tried that so I don't know Yes, so that was that's a very good question so I basically Okay, so you can import your DS output Yeah, so my next slide is actually actually this So I have a couple I actually tried this on Tableau as well but a couple of things that sprang into my mind is that number one you can use power BI kind of like for free and You don't need to use Tableau because Tableau like you have to pay after 14 days, right? So And then number two The setup for Tableau is actually slightly more complicated. You need to launch the server within power BI within Here, where's it help and you actually need to launch a server in order for you to start running Python calculations within Tableau and then after that a lot of your Python scripts are actually used as a form of Calculated field like a measure. So by and large even if you Are able to create the calculations like, you know correlation or whatever, right? You can by and large you are still restricted to the visualizations within Tableau. So You kind of need to Select the different so small You know like on the top right corner. There's the select different visualization Option and then you need to you can select different like violent plots. You can select box and whisker plots. So I Think if you are a Fresh Tableau user you kind of like, you know, you have to relearn the whole thing. So for me My objective is I have a Python script. It gives me and It gives me a visualization. I want to just park it into somewhere. I don't need to worry about all these things I don't need to worry about Vega D3. I don't need to worry about Altair and pop. I can get a visualization out pop I can have someone interact with the line charts the country data You know just zoom in zoom in out so that your managers can quickly understand the data So that's where I'm coming from. I think this you have the ability to run like really long machine learning scripts on Tableau but I haven't done much of that as well and like My main focus is just quick and easy quick and easy, you know Something that even a beginner matplotlet C-Born user can use cool Yeah, so I Actually wanted to demo but I'll just give you guys a quick View of how simple this can be done I'll just do a quick one. All right. So this is a fresh page, right and usually we get data from Excel So I'll just you know get from Excel and then you can get this thing and You can just load it just So then you know just Enable the script. I mean like I have I will I can post the link on how to install and run these things But I'll just showing you after after you in after you installed it So if I want to run some basic calculations like the pairplot just now Discount profit shipping cost unit price sales So all these things will be stored in a pandas data frame here and then What is required is? just To avoid issues I just drop NA data set equals data set Don't drop and a import Plot lip Plot SPL team port C born SNS SNS dot pairplot data set Plot dot show. I hope I didn't make any mistakes there So that's it. And then you know if you have other stuff you can just create a line chart and then sales by Whatever right like region maybe and then you can even have a slicer just something maybe like a province and Then like the person can just go in and interact and then this thing see the loading thing And after a while like I mean you can think of you can just basically put anything you want and you know it's a lot more effective and Yeah, let me just go back All right Of the data set that I just parked in oh, it's oh sure sure sure sorry about So The data set that I used is actually a very Common sales data set. Let me just hide everything You know, it's just Transactions or order date the usual stuff that you get within a business environment, you know Sills discount ship mode customer name product category Yeah, so it's very applicable very relevant to I think you know even everyday business analysts Yes, you someone else had any question Yeah, so for now it is hundred and fifty thousand rows Yeah, but like I said if you want if your intention is to feed in like an actual raw data and Perform like, you know a lot of data wrangling data cleaning on it. I wouldn't advise that I would advise you to plug in like final kind of like Prediction table so that you can very quickly run your heat maps and pair plots on it Yes, yeah, I say so you just need to click refresh Yeah This thing so the dashboard you can save it up in power bi.com But if you want to refresh the files have to because this is a desktop environment, right? So your excel files, it will remember where the excel file is every time you update the excel file Then it will just pull from the excel file. Yeah, it's not meant to be like a corporate level kind of visualization It's just very quick and simple, you know, very hacky. It's very I think useful for someone who's a beginner or intermediate Python learner Yes, I personally I use the pie the power bi dashboard to discuss like sales and Like all the purchasing decisions with my team It's not I don't I mean it's not here, but I basically run a lot of pythons Python scripts using a doing a get query from our from our from our Like trade gecko Shopify Post systems, and then I just perform all the calculations Either through Python or through power bi and then I use I show it to my colleagues for them to discuss Yeah, so you can just click publish here and then it will go up but because this is not a pro license so the visualizations won't show I mean just Everything else Yes So Python so power bi has a like their own like Software called a data gateway in which you will like run as long as the PC is running to query files From this particular machine so that you can constantly update and show the latest results as long as your machine is running So is that does that answer your question? Mm-hmm I am so okay, let me just show you what I would do. It's a it's a little bit hacky. So One sec All right, so because this is not a pro license these things don't show But if you have a pro license, I think you'll show so yeah, all right, so I saw someone raising a hand No, okay. So yep, basically, that's just a very specific use case for people who are just you know starting out to learn Python, you know, if you need to just if you if you deal with a lot of business analysis work You know, this is a very simple way for you to just visualize data, you know show some of your work On machine on machine learning data science incorporate them into your daily analysis works So I think that will really help you practice and it will really, you know Add value to your company as well. Hopefully because it has added for me. Yep. So I think that's it