1 00:00:00,090 --> 00:00:02,830 We talked in the last lecture about histograms. 2 00:00:02,850 --> 00:00:08,970 Now we want to talk about a very similar type of plot, which is the bar plot and an example of a bar 3 00:00:08,970 --> 00:00:11,670 plot or bar chart looks like this. 4 00:00:11,700 --> 00:00:18,960 Remember that we use histograms for continuous variables like age or time or temperature. 5 00:00:18,960 --> 00:00:22,350 So we have that continuous variable along the horizontal axis. 6 00:00:22,350 --> 00:00:28,380 When we use a bar plot, it's usually for a categorical variable which we put along the horizontal axis. 7 00:00:28,380 --> 00:00:33,900 So here, for example, is a bar plot of the number of times that each continent has hosted the Summer 8 00:00:33,900 --> 00:00:34,560 Olympic Games. 9 00:00:34,560 --> 00:00:41,070 So we have Europe, North America, Asia, Australia and South America and this variable which we could 10 00:00:41,070 --> 00:00:42,330 call continent. 11 00:00:42,330 --> 00:00:49,260 So if we say that along the horizontal axis here, we plot continent, continent is a categorical, 12 00:00:49,260 --> 00:00:51,360 not continuous variable. 13 00:00:51,390 --> 00:00:55,020 Obviously there's nothing continuous about different continents. 14 00:00:55,020 --> 00:01:00,960 There's no continent in between Europe or North America or a continent in between North America and 15 00:01:00,960 --> 00:01:01,620 Asia. 16 00:01:01,620 --> 00:01:08,160 In the same way that if we plot age along the horizontal axis, we can think about in between values, 17 00:01:08,160 --> 00:01:14,730 because age can take on every value as long as we can measure age to a detailed enough degree. 18 00:01:14,730 --> 00:01:17,460 But there's no in between concept here with continents. 19 00:01:17,460 --> 00:01:20,400 These are just different buckets, different categories. 20 00:01:20,400 --> 00:01:23,940 So this is a categorical, not continuous variable. 21 00:01:23,940 --> 00:01:30,600 When we have a variable like this, it's really common to use a bar plot or bar chart because it's one 22 00:01:30,600 --> 00:01:34,710 of the simplest ways to summarize and graphically represent. 23 00:01:34,710 --> 00:01:40,860 A set of data will usually just put the categorical variable along the horizontal axis and then we'll 24 00:01:40,860 --> 00:01:43,380 put count along the vertical axis. 25 00:01:43,380 --> 00:01:45,960 So we could say here that this is count. 26 00:01:45,960 --> 00:01:50,940 And again, this particular bar chart is the number of times each continent has hosted the Summer Olympic 27 00:01:50,940 --> 00:01:51,510 Games. 28 00:01:51,510 --> 00:01:58,410 When we're talking about a categorical variable, we usually refer to each category as an individual. 29 00:01:58,410 --> 00:02:04,010 So we plot the individuals of the categorical variable along the horizontal axis. 30 00:02:04,020 --> 00:02:09,210 Keep in mind, though, that for bar charts or bar graphs, not only can we create a chart where the 31 00:02:09,210 --> 00:02:15,270 bars are shown vertically, we can also create this same chart where the bars are shown horizontally. 32 00:02:15,270 --> 00:02:18,900 This is exactly the same chart, except that we have flipped the axes. 33 00:02:18,900 --> 00:02:24,840 We have the continents here, the categorical variable along this vertical axis, and then along the 34 00:02:24,840 --> 00:02:26,970 horizontal axis we have the count. 35 00:02:27,000 --> 00:02:30,330 Notice one thing that these charts have in common. 36 00:02:30,330 --> 00:02:37,650 We don't always have to do this, but it can be nice for a bar graph to organize the data from least 37 00:02:37,650 --> 00:02:43,350 to greatest or greatest, to least notice that we have arranged the individuals in this categorical 38 00:02:43,350 --> 00:02:43,860 variable. 39 00:02:43,860 --> 00:02:48,990 We have arranged the continents in such a way that we're showing the continent that's hosted the Summer 40 00:02:48,990 --> 00:02:54,480 Olympics the most times first, and then all the way down to the least number of times last. 41 00:02:54,480 --> 00:02:57,480 And then same thing here with the bar chart flipped the other way. 42 00:02:57,480 --> 00:03:02,520 We have the continent with the highest count on top, all the way down to the lowest count on the bottom. 43 00:03:02,640 --> 00:03:09,150 It can be nice to arrange the data that way because visually it makes it really easy for us to see highest 44 00:03:09,150 --> 00:03:12,030 count, lowest count and everything in between. 45 00:03:12,030 --> 00:03:17,580 But depending on the variable, sometimes it's not going to make sense to put the data in order by count. 46 00:03:17,580 --> 00:03:23,370 For instance, maybe our categorical variable is months of the year, so January, February, March, 47 00:03:23,370 --> 00:03:25,530 April, etc. all the way to December. 48 00:03:25,530 --> 00:03:32,400 If that's the case, we could certainly rearrange the months based on count and show the greatest count 49 00:03:32,400 --> 00:03:34,830 first and the lowest count last. 50 00:03:34,830 --> 00:03:39,870 But it also might make the most sense just to show the months in order, starting with January ending 51 00:03:39,870 --> 00:03:40,620 in December. 52 00:03:40,620 --> 00:03:46,950 It always comes back to what we're trying to represent and what makes most sense to communicate the 53 00:03:46,950 --> 00:03:48,600 picture we're trying to get across. 54 00:03:48,630 --> 00:03:55,980 Notice also that with these bar graphs, we always show a space or a gap between each bar, whereas 55 00:03:55,980 --> 00:03:59,760 with the histogram we eliminated that space between the bars. 56 00:03:59,760 --> 00:04:05,490 And so we had what looked like one continuous distribution that was intentional to indicate that we 57 00:04:05,490 --> 00:04:07,380 were dealing with a continuous variable. 58 00:04:07,410 --> 00:04:14,100 These gaps or breaks or this separation in between each bar indicates or gives us another hint that 59 00:04:14,100 --> 00:04:16,790 we're dealing with a categorical variable. 60 00:04:17,010 --> 00:04:23,430 Realize here that any bar plot or bar chart is associated with a frequency table. 61 00:04:23,430 --> 00:04:31,110 So the raw data for these charts might be the complete historical list of the host city for every summer 62 00:04:31,110 --> 00:04:32,100 Olympic Games. 63 00:04:32,100 --> 00:04:40,380 And using that raw data, we could make a frequency table so we could take every instance of the Summer 64 00:04:40,380 --> 00:04:45,420 Olympic Games and create a table where we group those games together by continent. 65 00:04:45,420 --> 00:04:47,270 And so our table might look like this. 66 00:04:47,280 --> 00:04:49,080 Europe, North America. 67 00:04:49,890 --> 00:04:54,990 Asia, Australia and South America. 68 00:04:55,170 --> 00:04:58,170 And then our count. 69 00:04:59,150 --> 00:05:05,810 Of games would be 16, six, three, two and one. 70 00:05:05,810 --> 00:05:07,610 So this is continent. 71 00:05:08,570 --> 00:05:15,890 And this is count and this is what we call the frequency table that we can use to build a bar chart 72 00:05:15,890 --> 00:05:17,480 or bar plot or bar graph. 73 00:05:17,510 --> 00:05:24,560 Obviously, once we have this frequency table, we would list the individuals in the categorical variable 74 00:05:24,590 --> 00:05:30,110 across the horizontal axis, or if we're doing our chart horizontally across the vertical axis. 75 00:05:30,110 --> 00:05:36,290 And then we would just make sure that each bar extends to the correct count of that individual. 76 00:05:36,290 --> 00:05:36,810 So. 77 00:05:36,830 --> 00:05:39,430 Europe has hosted the Summer Olympic Games 16 times. 78 00:05:39,440 --> 00:05:47,870 So when we create a category here for Europe, we show a bar that extends up to this point here, 16 79 00:05:48,080 --> 00:05:53,450 and then we show North America right about here at six, etc.. 80 00:05:53,450 --> 00:05:55,490 And so we plot those bars. 81 00:05:55,670 --> 00:06:01,400 And then the last thing we want to say about bar plots is that they are also great for comparing two 82 00:06:01,400 --> 00:06:02,600 different variables. 83 00:06:02,600 --> 00:06:08,750 For instance, maybe we want to compare summer and Winter Olympic Games in this chart we're showing. 84 00:06:09,550 --> 00:06:13,630 Summer games in blue and we're showing. 85 00:06:14,240 --> 00:06:16,510 Winter Games in green. 86 00:06:16,520 --> 00:06:22,280 If this is the case, we obviously want to make sure to include a legend like this that we've shown 87 00:06:22,280 --> 00:06:22,500 here. 88 00:06:22,520 --> 00:06:28,610 Maybe we would do something like this, but we want to indicate which series is associated with which 89 00:06:28,610 --> 00:06:30,020 set of games. 90 00:06:30,020 --> 00:06:36,590 And these bar charts are a really good way to show the comparison between these, because not only can 91 00:06:36,590 --> 00:06:41,270 we see the comparison between games for each continent, so we can quickly see here that Europe has 92 00:06:41,270 --> 00:06:46,190 hosted the Summer Olympics more than the Winter Olympics, whereas it looks like North America has hosted 93 00:06:46,190 --> 00:06:48,740 an equal number of summer and Winter Olympic Games. 94 00:06:48,740 --> 00:06:54,920 But we can also quickly compare the Summer Olympic Games by continent and the Winter Olympic Games by 95 00:06:54,920 --> 00:06:55,430 continent. 96 00:06:55,430 --> 00:07:00,200 So there's several things we can do all in one chart and we can do it very quickly. 97 00:07:00,200 --> 00:07:03,440 So simple idea here with bar graphs. 98 00:07:03,440 --> 00:07:08,600 The biggest takeaway being that we use a bar graph when we're dealing with a categorical variable, 99 00:07:08,600 --> 00:07:12,530 whereas we use a histogram when we're dealing with a continuous variable. 100 00:07:12,530 --> 00:07:18,620 And if we're creating a bar graph, we want to make sure to leave space in between the bars. 101 00:07:18,620 --> 00:07:24,440 Whereas when we create the histogram, we want to make sure to show no space at all between the bars 102 00:07:24,440 --> 00:07:29,240 so that we can make it clear whether we're dealing with a histogram or a bar graph. 103 00:07:29,240 --> 00:07:35,420 And then assuming we are dealing with a bar graph, if we're starting from a frequency table like this 104 00:07:35,420 --> 00:07:42,620 one, we just want to decide whether or not it makes more sense to order our data by count like we did 105 00:07:42,620 --> 00:07:49,040 here from largest to smallest count, or whether it makes sense to show the individuals in the categorical 106 00:07:49,040 --> 00:07:51,860 variable in a different order like we talked about earlier. 107 00:07:51,860 --> 00:07:57,170 If our individuals are months of the year and maybe it makes more sense to show them chronologically 108 00:07:57,170 --> 00:08:05,960 is January, February, March, etc., even if the counts then aren't in order from largest to smallest.