StatQuest: Histograms, Clearly Explained — Transcript

Full transcript

1. 0:00My cat
2. 0:02does stats
3. 0:04when she sleeps
4. 0:08I like to do stats how bout you when I'm awake
5. 0:15StatQuest
6. 0:18Hello and welcome to StatQuest!!!
7. 0:21StatQuest is brought to you by the friendly folks in the genetics department at the University of North Carolina at Chapel Hill
8. 0:29Today we're going to be talking front histograms, and they're going to be clearly explained
9. 0:35Imagine we went out and measured someone and they were this tall.
10. 0:40And then we measured someone else
11. 0:43And then we measured a whole bunch of people
12. 0:46We've measured so many people that the dots overlap; some dots are completely hidden.
13. 0:52We could try to make it easier to see the hidden measurements by stacking any that are exactly the same
14. 1:00But measurements that are the exact same are rare and a lot of the hidden measurements are still hidden.
15. 1:06So, instead of stacking measurements that are the exact same we divide the range of values into bins
16. 1:15And stack the measurements that fall in the same bin
17. 1:19This, my friends, is a histogram
18. 1:22Bam
19. 1:24The taller the stack within a bin the more measurements we made that fall into that bin.
20. 1:31Duh
21. 1:33We can use the histogram to predict the probability of getting future measurements.
22. 1:39I would be willing to bet that the next measurement we make is somewhere in this range.
23. 1:45Measurements out here are rarer and less likely to happen in the future.
24. 1:50If you want to use a distribution to approximate your data or future measurements,
25. 1:56histograms are a good way to justify your decision.
26. 1:59By the way, if you don't know what a distribution is, there's a StatQuest for that.
27. 2:05In this case we might use a normal distribution to approximate the data and future measurements.
28. 2:12If the data look like this
29. 2:15we might use an exponential distribution to approximate this data and future measurements
30. 2:22Note:
31. 2:24Figuring out how wide to make the bins is tricky
32. 2:28If the bins are too narrow, then they are not much help
33. 2:32In this case the bins are so narrow that pretty much every measurement gets its own bin
34. 2:37This doesn't give us much more insight than what we had before
35. 2:41so it's not very useful
36. 2:44And if the bins are too wide they are not much help
37. 2:48In this case the bins are so wide that the measurements are split 50/50
38. 2:54All this tells us this how many measurements are above the average and how many are below
39. 2:59this is more insight than before, but we can do better
40. 3:04Sometimes you have to try a bunch of different bin widths before you get a clear picture
41. 3:10In other words don't rely on the default setting of whatever program you're using to draw the histogram
42. 3:16You've got to try a bunch of different settings before you're sure that you've got the best histogram you can draw
43. 3:23Hooray, we've made it to the end of another exciting StatQuest!!! If you like this StatQuest and want to see more like it
44. 3:30Please subscribe
45. 3:32It's really easy, and if you have any suggestions for future StatQuests
46. 3:36Just let me know in the comments below. Until next time... Quest On!!!