AI-powered data visualization with Copilot

- [Instructor] Data visualization is vital for simplifying complex data, uncovering patterns, and effectively conveying findings to others. With so many chart options available, however, it can be challenging to determine the best starting point. Fortunately, Copilot can provide assistance by offering suggestions that inspire deeper analysis and improve visual representations. Consider the exercise file 03_03_data_visualization, which again contains data on the 15 largest cities in the United States. We'll open Copilot, and we'll get right to our own prompting in-app skills here. I'll begin by graphing the population of each city, and I will be kind of deliberately vague here. Let's say, "Plot the population of each city." I'm also going to include, "Don't use Python" just because I know, lately, that Python has been used a lot with the prompts I've been getting back. Okay, and we do get a plot here. We can add this if we wish. And if we wanted this sorted, that might be a question, again, try to ask that here. I'll go ahead and try, "Can you sort this chart by high to low?" And ideally, Copilot will understand what I mean here. I guess I could have said, "Sort by population from high to low." Okay, it's going to give us a formula. We could try that, or I might ask something like, "How do you sort the values of an Excel bar chart from high to low?" So, depending on what Copilot is and isn't able to do for you with the output, you can get some additional assistance by some more generic kind of prompting like this. Okay, so let's try another one. We'll experiment with an aggregation this time. So, I'm going to ask for the average area by its capital. Okay, I'm also going to specify, "Don't use Python," and we'll see what we get here. And we do get a chart. I'll go ahead and insert this one. I'm a little happier with it than the population chart. At least, it's well sorted. So, let's add this to a new sheet and see what we get. And this is a case where if you did want to maybe round these in the pivot table, would be worth maybe re-prompting or asking Copilot just generically, "How would I do this?" if you don't know how to do that yourself. But that's looking pretty good. And you're noticing that Copilot is favoring bar charts in these situations. That does demonstrate an important principle that having that urge to just innovate with new graphs for novelty's sake can be strong. It's often not necessary. Bar charts are a standard choice because of their simplicity and their effectiveness. With that being said, let's examine a situation where a bar chart might not be the best choice, such as when illustrating the relationship between two quantitative variables. So if I go back to my dataset, we've got area and we've got population. What I want to do with this chart is visualize the relationship between the two. And we'll say, again, "Don't use Python." We'll see what Copilot will suggest here. Okay, now we do get a scatter plot. If you're not familiar with scatter plots, you could always ask Copilot for more information. You know, "How do I understand this? What is a scatter plot? What are the pros and cons?" I'm going to try to modify this, "Can you add a trend line to this plot?" And we'll see if Copilot can comply. If not, just add the sheet to Excel. Okay, right, so let's go ahead and if we can, this Add to sheet button may or may not come back for us. So, you'll see here that Copilot isn't able to do it for us, but it will kind of compromise and give us the instruction. So I'll come back here, I'll Add to sheet. Okay, and then, if we wanted to continue adding the trend line, we could do that using the instructions given by Copilot. In summary, Copilot facilitates data visualization by suggesting efficient and straightforward chart types like bar charts and scatter plots. It emphasizes that utility often trumps the need for novelty, leaving room for users to refine and personalize their visualizations.