Sure! A Q-Q plot allows you to see how well a collection of data points fits a particular probability distribution. For example, let’s say you generate 20,000 random points from a student T distribution, and try to fit them to a normal distribution with an identical mean. The Q-Q plot may show that your points diverge from the distribution expected if the data was drawn from a normal distribution, but that it is a good fit for a student T distribution. I’ve found (in my limited coursework experience) that you often need lots of data points to get clarity on this, because the differences are most apparent in the extreme values far out in the tails.
This is connected to a broader class of mathematical challenges, such as curve fitting. If you look up courses on predictive analytics, you’ll find many examples of challenges related to the one you articulate here, along with tools for solving them!
Sure! A Q-Q plot allows you to see how well a collection of data points fits a particular probability distribution. For example, let’s say you generate 20,000 random points from a student T distribution, and try to fit them to a normal distribution with an identical mean. The Q-Q plot may show that your points diverge from the distribution expected if the data was drawn from a normal distribution, but that it is a good fit for a student T distribution. I’ve found (in my limited coursework experience) that you often need lots of data points to get clarity on this, because the differences are most apparent in the extreme values far out in the tails.
This is connected to a broader class of mathematical challenges, such as curve fitting. If you look up courses on predictive analytics, you’ll find many examples of challenges related to the one you articulate here, along with tools for solving them!
Yes, you can get a very nice visualization with a simple one-liner in R! I use qqPlot from the cars library.