Let's take maybe step by step. Let me show you some of the things that you can do. Even I guess what I could do is put inside of here at the very beginning, I can put a section called Resources. Let's do this. We'll go through here and we'll call this Resources, so Business Intelligence/ Data Science Resources for Low Code/No Code. I guess we can use our notebook to take notes here. Step 1 would be let's look at Apple here, maybe I'll just do a section here. This is Apple. What do they have that's interesting? I think one of the things that's very interesting about Apple is they have something called Create ML. Create ML allows you to create machine learning models with low code or almost no code. Here's a picture of this is that they have a system that will accept images and allow you to create a model. How does this work? This is really the feedback loop here, is you ingest the data like I talked about. You do the training and you do an evaluation. It's the same type of workflow. Let me just put a note here in this notebook. We can say Create ML and put a link here like this. We got Create ML there. Why do we care about Create ML especially if you're into a low code no code solution? Well, let's look at this. Creating an Image Classifier Model. What happens with an image classifier, we know high level, you can train a classifier to give it many examples like an elephant, a giraffe, whatever, and tell it to recognize the differences between those images. What's nice about this approach here is they only need about 10 images but a little bit more could be helpful per category. You put your data into a folder just like this and then once you've collected that data, and maybe even I can take a picture of this and we'll see a copy image address. Notice that in my report, which is kind of neat, is I can click this link. Actually click this. I can actually make this into a picture that if I was doing research for a company, I think it's like this. Actually, let me see it. I thought it was this. There was a way to do the link of a picture here, but I think it's something like this maybe. Is that how you do it? What is it? Like this? If I for some reason I can't figure out, I'll just do this. Let's see Markdown Image Link. What is it again you do? I'm so lazy, I can't even put this in here. I'll just do this and then I will put in the path to that image here which is this I believe. Does that work? Anyway, I'll fix it later. We'll just keep this link in here for now. But basically, the core idea is that you want to organize your testing data into a directory structure, put it into a project like this, image classification, choose the options for your new project, and then go through and drag that into this GUI. Again, no code at all. There's no code that you're running, you're just creating your images into a directory dropping in here. Then you have in this particular situation, the ability to add noise, blur cropping, et cetera. You then have some testing data. This after you train it, you want to make sure that works right, so you put the testing data inside. After you're done with this, you can, again, as I mentioned, add some parameters to maybe, enhance the data. A lot of times when you're doing a computer vision problem, you don't have enough data. You can select augmentation, which will allow you to get more data because you can flip the image, and rotate it and do things like that. Once you train it, it'll go through and give you results. We don't have time to get into all the details here. This is something I would recommend, trying on your own. But the core idea here is that you can use, no code-no code tool and come up with something, that helps you do your business intelligence report. Maybe it's that you want to analyze competitor's product images, and you want to train a classifier to detect any time, you see competing products. You can go to a store and you train a classifier, and you could take pictures of the products and then you see, this is a competitor's product is in the store. We need to investigate with this competitor's product is, as just one example. That we'll just go through this step by step. We got Apple here. That was one. Who's next here? Let's do this for links here. Then we'll add another company. I think we had GCP. Let's go to GCP next here. Under GCP, it's nice Google, there's a lot of powerful stuff that you can use. Let's start here with h GCP console. I think probably auto-ml be one of the more interesting ones. There's two forms of AutoML that I think business intelligence people, would want to know about. The first one would be AutoML Vision. Here's an example of it. The way it works is that, you upload data into their cloud platform, in this case, I would say new data set, and you decide, just like the Apple example, I want to predict a particular label like one label, or I want to do multilabel classification, predict all the correct labels, or I want to do object detection. Let's select that, for example, all you need to do is either upload the images or select a CSF file, that points to those images that you put in cloud storage. You just, again, don't have to write any code. Just drop it into the cloud storage, say what the name of that file is. It's a cat. It's a dog. Whatever the label is that you want to detect. Let me make this a little bit bigger. Once you do that, you go your data set, and we go back here. It will show up. It'll be visualized. You can look through and filter and see the different pieces, of data and analyze them, and prepare your report. Then you click on this train button and again, no code, just click "Train model.' It'll train it and then it'll come up with, a result that looks like this where you will get a bunch of metrics, that show you even if you're not an expert in machine learning what's happening. This is fairly easy to understand, this confusion matrix is that it shows you when it was correct. If it's in blue, that's its accuracy. Then these other sections are when it was misclassified. Right. Again, I can even look at each of these, and it will visually show me when it was correct, the true positive, when it was false negative. It should have predicted it, but it didn't. Then false positives. Then later, if I wanted to be also a little code, no code, I could export this model, and make it run on a mobile device. Maybe I'll just put a link to that here, and we'll say AutoML Vision. There we go. We got a link to this. What else can you do with the GCP platform, that does low code, no code? I've done this with MBA students. AutoML tables is another one, that's pretty interesting. You can take tabular data, which is something you'll get all the time, as a business professional. With tabular data you can go through, and upload it as well. I'll just show you real quick how I would upload it. It's a new data set. We create a new data set. You could either import it from, the Google big query system, which I'll show you briefly, which allows you to do ad hoc queries, on data inside of a console. You can upload it from a CSE file, in cloud storage, or just take a file off the computer, upload it. What does it do? Well, this is, I think, the future of machine learning, and why I think it is, it' to not be an expert in software engineering. If we look at a data set like this, you ingest this data and it shows you, all of the key things that you would need. Let's go back. Here we go. If I go through here, it shows me all of the attributes that are in my data sets of all the columns in a spreadsheet. It will infer the type correctly. Auto mail will go through here and find out the numerical types, the categorical types, the text types. It will even show through, basically descriptive statistics like how many values are unique, et cetera, and also if I click here, I think I can scroll over a little bit. That it will also show you the correlation. This is something that is called feature engineering. I can see that certain types have higher correlation with the target that I'm going to try to predict. Really, the only thing I need to do is maybe look at this to make sure that everything is correct in terms of the data types and then also select what I want to predicts. I want to predict whether someone will be married or not, or I want to predict what their income level is, or whatever I'm trying to predict. Once I predict this, then I go to evaluation and I look at the metrics here and I can see exactly the accuracy. In this case, this is 83 percent accuracy or area under the curve for precision recall curve, and then I can also go far down as well, look at the confusion metric, what was a good at predicting. It's pretty good at predicting whether you won't make a lot of money. It's okay, but is less good at predicting whether you will be greater than 50,000. But we can see here the future importance. This is, again, for business professional, yeah, maybe what I do here is I just do this, I take a picture of this. Actually just let me download this. Well actually, yeah, that's a pretty interesting. I can just download this and then I could put this in to my report. Let's go through here and in fact, again, I don't even have to write any code, I just see add file, upload some of my code here feature importance. Let's put this inside of here to make this change. Now I can put this link, and if you click on this, you can see the different features. Even if I wanted to, could cheat a little bit or write a little bit of code. I could ingestion in here, I could say import pandas spd and say df is equal to pd.read_csv and put that path in and then say df.head, I don't know, 20, I don't know how many rows there are. I've actually got a little bit of raw data that I could look at in my report. It looks a little bit technical. Also by the way, if I go back to this I could, this is what I was going to show you earlier. If I do a screencap, I could actually do this. I could put this into my report without writing any code. Just take a screen capture, go to this project here, create an issue and say new issue images. Then I'd like to just drag in a screenshot like this, and it creates the URL for me. That's what was missing. I'm so lazy that I don't want to do any work. I actually like no code low code, and here we go. I go to this in then maybe I put this under my exploratory data analysis picture. Here we go. We got some exploratory data analysis, and then I say Return, Shift Return. Now we actually have a nice graph, and I didn't even make it. Somebody else made the graph for me, and I have a good feature analysis, and I could say here, these are some important features. You can see this raw data in adjust section. I'm building up this portfolio and then again, if I want to save this and I want to let other people see this, I can just go again to save a copy in GitHub, go back to my report here and put it into business report and check it in, and also if I want to give you a copy to all this, you can get a copy as well. I'll go through here and put this in to chat as well and just throw this in. There we go. Actually we should do this whole repo do this. Now that I've got this whole report created, all the technical details hidden in here, and maybe later I'll tease this out. But let's keep moving. This is great, I can do this, with DCP auto-mail, but what else can I do? What are some other companies, that I can use their tools? Well, if we look back here, what do we have next? We have AWS. Let's go to AWS next year. Let's take some of their skills. AWS has a lot of stuff that you can use, and it's amazing, actually how many services they have. One of the first ones that we can log into here, with is potentially, let's start with NLP. With NLP the main service, that you can use, is this comprehend service, and if I click on ''launch comprehend'', you can see, we'll be able to do some NLP, which I'll do in a second. But I also wanted to show, a different comprehend. There's two forms of AWS comprehend. There's text-based, and there's also comprehend medical. Depending on what problem you're solving, you could launch a different one. The reason I want to show this is, let's say you're working for a medical company. You could actually just paste a medical record, for somebody in your company. Again, right now writing any code. Here's one. This is their medical record like this, and we say analyze. Inside of this, you can see, all of the different text, so visually now we've created a report, and I can take this report, and I can give it to the doctor or to maybe the CFO of the company, and say, we have some really important facts here, that we need to analyze. What I could do, is maybe minimize this, and take a picture, for example, of this particular output. Let's see here, what's a good one to show. Let's take this. Here's a good chart, and I could again take a screenshot, and maybe. I think this is just good. Screenshot there good. Got the screenshot, and we'll go get back to this repo, go to my issue here, go to images, and put another one inside. We'll take this new screenshot, and we'll throw that in. Great. Now we have some details, that we can put into our technical report. We have another detail here. Let's go through here. We'll say maybe another section, and we'll say this is a medical record, paste that in. Now we've got another chart. But let's go back to, the regular comprehend real quick, and again one of the main ways to probably do this would be to let's say, I was going to do research on a competitor. This is a very business-centric approach. I would probably start with Wikipedia is a good source for doing research, and let's say it was Nike. I want to know what, let's say I'm a shoe manufacturer and I want to see, what I can learn very quickly, so I can give this report to somebody in my company. I'm going to take maybe the first section here, and I'm going to copy it, and then I will go to this input text here, and just paste it in just like this. I'm going to say analyze the text. What it will show me is all of the entities inside. But I also can go through, and find key phrases, and I also could find syntax or sentiment, and we can see that, in fact, there may be one thing that I would want to look, at is the sentiment. In fact, what could be important here would be that if it's a journal article, it should probably be neutral. But we see that this is actually a little bit positive, which is interesting. I could then maybe look further, and say I want to do some research. What if I said I look for news about Nike? I'd go to, let's say Google News, or something like this, and I type in Nike. Let's look at Nike here, and we find some news about Nike, and I want to see what my competitor is doing, and I want to see if people are writing good things about them, or bad things. Here we go I'll copy this, and I go back to this business analytics section, or comprehend analyze this, then I look at the sentiment, and I see that also this is neutral. It looks like a lot of people are writing neutral articles about it. But maybe, I can find another article. Here's another one. Let's just analyze a few and I can do this very quickly, because I'm not writing code. I can do this very quickly. I just cut and paste and I just do some very quick research. Let's see here, this looks like a good one to analyze, sentiment here seem pretty neutral. It looks like maybe this competitor people aren't writing very very positive or very very negative. They're more of a neutral one and maybe even I just go like this and I say I did a little bit of analysis and it looks like, a lot of content around Nike is pretty neutral. That's my initial findings for using these off-the-shelf p tools. Again, go through here, let's add another section, it will say, Nike research, and paste this in. But what else can you do with the AWS? There's a lot of other stuff we can do with AWS. One thing you can do is they have some computer vision tools. One of them, that's pretty interesting is recognition, and with recognition, you can actually do lots of different computer vision tasks without needing to write any code at all. Here's one that could be pretty useful for let's say we're still analyzing Nike is if I go to text and images here, what I could do is I could go back to this article and I could just see again, maybe I want to start looking at what color of things my competitor is putting into their product, or I can go here and I can save a picture of this. I can go to recognition, I can upload it, put it into this section, and here we go, so I didn't know this, I couldn't read that. Now I know this in fact, has the word Fease so maybe that's a new feature. Maybe this is something that we should start looking into. In fact, I can grab this if I wanted to or this is part of the English version is probably easy enough and maybe even I take a picture of the whole thing. I can just take a picture of this and just say, look, we're detecting that Nike has something called Fease. Maybe we need to do a little more research on Nike, to see if we can figure out what they're up to. Let's add some more will say computer vision research for Nike, here we go and again take a picture of this. I think it was the screenshots here, let's double check. Actually, I can put it into the images, that's a better way to do it. That's too big it doesn't like that picture is too big, so I'll just make it a little bit smaller we'll just do this. Let's make it smaller and go back to images here, and there we go. Comment, put this in, there you go got some computer vision research for Nike. Refresh this. This Col-ab notebook is struggling is for some reason. Maybe what I'll do is go back here. Not sure why this thing has got problems, exit. Here's my computer vision. Research and paste this in. We can do text recognition, label recognition, scene detection of all kind of stuff. Also I guess I can take that same picture and go through here and do research and see information about it. Maybe this is all I need to do, is just take a picture like this that says, here's what I found out. There are making a footwear it's a running shoe to sneaker, and it's got the word Fease in it or Flys or something like that. We go through here and do that. The other thing, though, that you can do, I think that's what I was screwing up is you put into. I think the problem is that I want to put this into the images section first, or else it gets too overloaded. You let GitHub do the hosting like this. That's probably good enough, I have enough of these pictures for my research. There you go. Now we have additional information. What else can AWS do? Another thing that AWS can do is it also has a tabular designs. We can go through here, and we can go to a service called Machine Learning. This Machine Learning service also allows you to ingest tabular data. The way this works is, you put the data into Amazon first, then you tell it to do an analysis. If I go to Amazon Machine Learning, and as I Create new Data source and ML model, paste this in. Here's a location to the data, I call this banking-data, and I verify it. It does that same process. It goes through, it looks at this data, then if I say yes for the column names in the first line, it will allow me to accept this data, then if I go to next, it'll ask me what do I want to predict, which is the whole point of putting it in here. If I go to the last section here, there's a target value, which is, I want to predict whether they deposit money in my bank or don't. This is banking customer data. If I click "Yes" and click "Continue", it'll prompt me to make some choices. I just keep clicking "Continue". It'll say basically generate a machine learning model that does a prediction whether these customers will deposit money, or not deposit money. I click "Review", it will go through, and it will change the model. That's it. I just upload some data, select the the columns, select the thing I want to predict, and it'll train the model. What does it look like at the end is, if I go to end result, is that, it will show me the metric, just like with Google, I also could look at the evaluation results and dig in deeper. We can see that this is area under the curve, I could explore the performance, do trade offs as well, save that threshold, then maybe give this to somebody else, with zero code to do this prediction. Let's go through another service on AWS. We also have QuickSight. What's great about QuickSight is that it also has this low code/no code feel to it, where you can just upload data sets. I think I have some data that I just downloaded. I'm going to say new data set, and I'm going to upload a file. Also, if I'm working at an existing company, maybe they have the data in Salesforce or they have it in Spark, or Jira, or GitHub, but in my case, I'll just upload the data, go to Feature Importance, take that data in here, it will give me a preview that shows me the columns, then what the values are, I go next, then I visualize. From here, I could then decide what kind of things I would want to chart. I think in this case, it will create some default visualization, and show me some interesting metrics. There we go. I have some metrics, and you can play around with coloring, and charting. Again, this gives me automatic insights, and it is something that I could then later share with other people in my company. I could click "Share" and I could publish a dashboard, or I also could print this, I could put this into a report, I think I can put this into a PDF, then link it into GitHub. What's nice about these BI tools here is that, again, it allows me to just upload the data that I pulled from some other source, and this data came from some other machine learning tool. I can have many low code/no code tools that I combine together, but for fun, let's just maybe even take a picture of this like this. I think this one also, you can say duplicate visual. I think there is another way to actually export it as a PDF. In a nutshell, those are some of the high level tools that you can use on AWS.