From the course: Learning Power BI Desktop

Create a visualization using Copilot

From the course: Learning Power BI Desktop

Start my 1-month free trial Buy for my team

Create a visualization using Copilot

“

- [Instructor] Here we are in Power BI Desktop, with the Copilot pane open. And we're being offered some prompts to create a new Report page, to suggest content for a new Report page to answer a question about the data. So, let's see how this works. But first, we're going to need some data because right now, we have no data. Click Excel Workbook on the Home tab, and go to the Chapter 07 folder in the Exercise Files. And open Catalog_Requests, which is a small dataset, 200 rows or so. And we're going to do one transformation on the data on the way in. In most of this course, we transformed the data after we had it in Power BI Desktop, but we're gonna actually transform it as part of the load event. So I'm going to choose Table 1. And I'm going to tell you that as soon as you see data about the United States, that has addresses in it, you'll know that if you have addresses in New England, that your Zipcodes are going to look like this. Some of them will be four digits, and some will be five because the New England Zipcodes start with a zero, and that's trimmed off because the Navigator believes that that's probably a whole number. So let's click Transform Data, and we'll throw ourselves right into Power Query directly. And slide across to Zipcode, this column. And let's go to Transform, change the Data Type to Text, rather than Whole Number. And Power Query tells us there's already an existing type conversion. Yes, it was the type conversion that changed text to a whole number. It's like, no, let's replace that. And the joy of doing this at this point is that because we're doing it as part of the load, it's really easy for us to get these Zipcodes right the first time. Let's go back to Home. Close and Apply. And we'll have some data shortly. Here's our Data, our columns. And if you'd like to see our data in the Table View, you can do that right here. So what we really have is data about geography of requests for catalogs, dates about requests for catalogs, and then finally the type of catalog that was selected because we have Sports, and Gifts, and so on. So, this all looks good. Let's go back to the Report canvas. And you'll notice that here, in the data view, we don't have access to Copilot. There's nothing for us to do with Copilot here. But when we go back to Report view, here's Copilot. So let's first ask a question about the data. I click, "Answer a question about the data," that prompt is sent to Copilot. Copilot replies with, "Great, what would you like to know? I can search on visuals, or I can create new visuals based on the underlying data." And so in the chat box at the bottom, I'm going to ask, "How many of each type of catalog was requested?" And click Send. Now, I know that that might be a little bit vague, what type, but Copilot is checking the underlying data and it is drafting a response. And here's the response, "Based on the available data, Sports catalogs were requested most frequently." Now, check this out. There actually isn't a particular column called type, there's one called Catalog Type. But Copilot was able to find that, to look across at the column heading and say, "One of these includes the word type, I bet that's what they wanted." Now, you can ask, "How did Copilot arrive at this?" It says here's the data used, there were no filters applied. That looks great. And if I want to add this to the page, I simply click, and I have a new visual that I created by having a conversation with Copilot.

Contents