Plotly’s Post

Transcript

And where I'm actually going to start today isn't with the traditional Dash workflow, which is a little bit more coding intensive. We'll get into that a little bit later for those of you who are already familiar with Dash and how it works. But I wanted to show an even simpler way to get started. And this is within something that we've been working really hard on over the past several months called App Studio. What App Studio allows you to do is take very simple code or even just a data set to start and then build an application from that using a variety of tools like UI and AI. So when I click that. It'll bring me to a new screen. Where if I want to make any modifications to the data or better understand the underlying data set. I have this summary view. I can also define custom filters and aggregation functions if I want to change the shape in this case. This is a pretty simple data set so we don't really need to do any manipulations but the options there if we want it. And then getting into the really interesting part, actually starting to build analytics on top of this. So you may notice on the right hand side we have this plot chart button and when I click that. That brings up a screen. Where I'm able to start building? Any sort of data visualization that I want, I'm going to go into my chat box at the bottom here I'm going to say generate a box plot. Showing prices by company from 2000 to 2005. As I'm typing this in, it's going to make a call out to that model. It's going to think for a moment, and then it's going to generate a pretty good visualization to start. Just like that line chart that we generated on the previous screen. This is fully interactive and I can see the underlying code to generate this. Notice how it's a little bit more complex than the previous example. Now I don't just have to take this graph as is, I can ask it to make modifications so I can say for example add a comparison. To price data by company. Between 2006 and 2010. Hopefully that's enough for it to understand what I'm asking, so I'm going to hit enter. And then again, it's gonna go ahead, it's going to do some thinking. Ohh, interesting. So in this case, it's not taking that context that I had before. It's actually generating a new graph. So AI isn't always going to know exactly what I'm asking it to do. So in this case, I'm going to say. You know, instead of generating a new chart. Create a box plot. Calling. Uh, company price data from 2000 to 2005 and another showing data from 2006 to 2010. That's the thing with large language models that's really nice is that you're able to iterate and without having to write any code, you can start to generate. These visualizations. Notice how we have this new section of our app at the top. Where we're not just. Viewing the graphs that we created, notice how that filtered data set that I added is in here now too, but we also have this dropdown at the top. And what that allows us to do. Is without having to write any code. I can just interact with. Those drop down values and that it'll filter. All of the components on the page that rely on that drop down. So maybe now that I have this filtered data set, I don't want to view full data set. I didn't see more clearly that as I. Make changes to this drop down. It's responding. To what I'm selecting. So in just a few short minutes, using large language models, using code, and using our UI, we're able to build a pretty sophisticated application. But everything that we built using this workflow and by the way. Even if there's stuff that doesn't directly affect the execution of the app. You know, like the summary of what the app is or the title. I can tweak those things as well. So I have this starting point for my application. I want to take this a little bit further. App Studio also gives you the ability. To generate all of the underlying code. Now, I understand this might be a little bit overwhelming to folks who haven't seen Dash before, or even to those who have played around with Dash before. It's a lot of code. But the really nice thing about all this code that you're seeing here is that I didn't need to write this myself. This is all stuff coming directly from that workflow.