1
00:00:00,270 --> 00:00:01,530
Welcome back.

2
00:00:01,540 --> 00:00:04,350
And so we've done a whole bunch of plotting so far.

3
00:00:04,560 --> 00:00:09,170
And don't worry if you're sort of a little bit confused at the moment like here like Daniel showing

4
00:00:09,170 --> 00:00:12,960
me all these plots I'm looking at different data sets.

5
00:00:13,020 --> 00:00:14,370
How am I going to remember all this.

6
00:00:14,700 --> 00:00:15,320
It's all right.

7
00:00:15,330 --> 00:00:20,370
That's perfectly fine to feel like that it's going to take a little bit of practice to start to understand

8
00:00:20,640 --> 00:00:25,760
different kinds of plots what you should use which plot for but that's all part of the process right

9
00:00:25,800 --> 00:00:30,950
is just having a little practice having a go and just trying it out seeing what it looks like.

10
00:00:30,960 --> 00:00:32,130
Does this make sense.

11
00:00:32,610 --> 00:00:37,590
So that's why in this video we're going to do a similar plot to what we've just seen before on the car

12
00:00:37,590 --> 00:00:39,980
sales data set on another data set.

13
00:00:41,370 --> 00:00:44,350
Let's try on another dataset.

14
00:00:44,370 --> 00:00:52,430
And if we come up here we've got heart disease dot CSB which we've seen before which is a data set which

15
00:00:52,430 --> 00:00:57,440
has different parameters of patients and whether or not they have heart disease.

16
00:00:57,440 --> 00:01:06,370
So let's import that as PD not read CSB as we're going to call it just heart disease and so we'll go.

17
00:01:06,380 --> 00:01:10,680
Heart disease starts see as very beautiful.

18
00:01:10,760 --> 00:01:16,340
Now we'll have a look at it just to make sure we know what's happening will only be the first five rows

19
00:01:16,340 --> 00:01:19,890
because about 300 and this one excellent.

20
00:01:19,980 --> 00:01:22,450
We have these parameters we've seen them before.

21
00:01:22,450 --> 00:01:24,650
I've got a target column right on the end.

22
00:01:24,690 --> 00:01:32,500
So what we might do is create a histogram just to see the distribution of the age column.

23
00:01:33,270 --> 00:01:43,760
So heart disease age dot plot dot hist and we might just see the default number of beans.

24
00:01:43,760 --> 00:01:46,820
Can you remember how many that will be shift tab.

25
00:01:46,820 --> 00:01:52,430
Beans is going to be 10 beautiful so we'll just leave it at that OK.

26
00:01:52,660 --> 00:01:55,250
So this is where we're starting to see that curve right.

27
00:01:55,270 --> 00:01:57,270
This is that normal distribution.

28
00:01:57,280 --> 00:02:02,740
It's very common so we're kind of getting that curve here.

29
00:02:02,750 --> 00:02:10,150
We would say that most of the patients are around it's this line here so maybe late 50s early 60s thereabouts

30
00:02:10,160 --> 00:02:12,500
and what if we changed the number of beans so let's go.

31
00:02:12,500 --> 00:02:16,280
Beans equals 20 OK.

32
00:02:16,280 --> 00:02:23,300
We're starting to lose that curve shape a little bit 30 still relatively there we're definitely starting

33
00:02:23,300 --> 00:02:30,920
to see this more pronounced line here right at the late 50s there jacking up to 50 and see what happens.

34
00:02:30,930 --> 00:02:37,290
All right so that shape is still relatively there our initial shape that what we got from just having

35
00:02:37,290 --> 00:02:42,930
10 bends is kind of consistent the whole way through and maybe we'd look at this and go Okay this one's

36
00:02:42,930 --> 00:02:44,300
a bit of an outlier.

37
00:02:44,310 --> 00:02:45,930
These are a bit of an outlier.

38
00:02:45,960 --> 00:02:47,010
Why are they outliers.

39
00:02:47,010 --> 00:02:49,640
Well because they're so far from the middle.

40
00:02:49,710 --> 00:02:54,750
When you hear the word outlier that's kind of what it's talking about in statistic terms.

41
00:02:54,750 --> 00:02:59,270
It's about more than three standard deviations away from the mean.

42
00:02:59,280 --> 00:03:03,900
Now of course that definition will change depending on what data are working with.

43
00:03:03,990 --> 00:03:06,480
But this is kind of a way that you can visualize it.

44
00:03:07,360 --> 00:03:09,340
So let's take that back to 10.

45
00:03:09,390 --> 00:03:10,830
That looks pretty good.

46
00:03:10,980 --> 00:03:11,780
All right.

47
00:03:11,850 --> 00:03:17,220
Now these kind of visualizations are the things you'll do whenever you deal with a new data set.

48
00:03:17,220 --> 00:03:18,000
Let's finish.

49
00:03:18,000 --> 00:03:24,760
We have there's one more type of plot we haven't looked at directly from a data frame that's subplots.

50
00:03:24,810 --> 00:03:26,490
So we'll finish that off.

51
00:03:26,490 --> 00:03:31,480
We'll go to heart disease or head OK.

52
00:03:31,480 --> 00:03:32,500
This is our data frame.

53
00:03:32,500 --> 00:03:36,850
Now let's plot every single columns histogram in one plot.

54
00:03:36,850 --> 00:03:38,400
Let's see how we would do that.

55
00:03:38,650 --> 00:03:40,260
Heart disease.

56
00:03:40,260 --> 00:03:42,090
So we've got the age histogram here.

57
00:03:42,120 --> 00:03:48,700
But let's just make one for all of these and check the distribution dot plot haste.

58
00:03:48,730 --> 00:03:53,950
So remember this is just going straight from the entire data frame up here we selected the age column

59
00:03:55,370 --> 00:04:04,660
dot haste and maybe we want subplots equals true that makes sense because we want to put it all on the

60
00:04:04,660 --> 00:04:07,530
one plot keyword arc.

61
00:04:07,540 --> 00:04:08,590
So that's part of that.

62
00:04:08,620 --> 00:04:13,400
Maybe if we could read down a little bit examples.

63
00:04:13,550 --> 00:04:19,430
So this is one of those scenarios where if you wanted to use this dot his method and have it all on

64
00:04:19,430 --> 00:04:26,090
one plot because the doc string doesn't really have an example of putting it all with the subplots.

65
00:04:26,120 --> 00:04:32,120
This would take a little bit of research so you might look up a question like how to plot a histogram

66
00:04:32,150 --> 00:04:36,970
on subplots from a panda's data frame and then you'd find this little parameter here.

67
00:04:37,010 --> 00:04:41,450
Only reason I know it is because I've had a little bit of experience plotting subplots from Panda's

68
00:04:41,450 --> 00:04:43,050
data frames.

69
00:04:43,050 --> 00:04:49,100
So let's see what this looks like might take a little while because we're plotting 10 columns or 15

70
00:04:49,100 --> 00:04:50,940
columns thereabouts.

71
00:04:51,000 --> 00:04:53,660
All that doesn't look very good does it.

72
00:04:53,730 --> 00:04:58,400
We're going to overlapping plots we've got legends everywhere we've got these things aren't even to

73
00:04:58,400 --> 00:05:04,740
scale so maybe to fix it up we'll change the fig size.

74
00:05:04,930 --> 00:05:08,490
So what do we want we want with height.

75
00:05:08,500 --> 00:05:13,660
We want it to be a bit higher so it stretches it stretches it out because there's hardly any space here

76
00:05:16,960 --> 00:05:17,270
OK.

77
00:05:17,290 --> 00:05:23,780
That's looking a little bit better but still you've got columns like this which are just all clumped

78
00:05:23,780 --> 00:05:29,900
up here because what it's doing is it's using the same scale for all of them but in reality these all

79
00:05:29,900 --> 00:05:31,010
have different measures.

80
00:05:31,010 --> 00:05:32,480
That's not very ideal right.

81
00:05:32,480 --> 00:05:38,390
It might work out for some like this one cholesterol might work out for that column but it doesn't work

82
00:05:38,510 --> 00:05:40,010
at all for the other ones right.

83
00:05:40,010 --> 00:05:41,740
So slope I'll peak.

84
00:05:41,750 --> 00:05:43,240
Exactly.

85
00:05:43,340 --> 00:05:44,830
So what can we do.

86
00:05:44,840 --> 00:05:49,030
Well this code runs but it's not ideal.

87
00:05:49,060 --> 00:05:54,900
Then there's different scales and it's probably a bit too much going on for this plot to even be meaningful.

88
00:05:55,250 --> 00:05:56,630
But this is a good segway though.

89
00:05:56,960 --> 00:06:00,790
Now we've covered a few simple plots directly on a panda's data frame.

90
00:06:01,340 --> 00:06:05,780
Let's have a look at how we'd make some visualizations using the O method.

91
00:06:05,800 --> 00:06:11,230
So remember back in the previous video we talked about oh method which is subplots.

92
00:06:11,390 --> 00:06:18,110
Don't worry if it's confusing we'll have a look in a second but what we've been using so far is plotting

93
00:06:18,110 --> 00:06:21,830
directly from pandas using this stateless method.

94
00:06:21,830 --> 00:06:23,840
So in the meantime I have a practice.

95
00:06:23,840 --> 00:06:25,040
Keep using the plot.

96
00:06:25,190 --> 00:06:26,270
Make another plot of your own.

97
00:06:26,270 --> 00:06:32,090
Before we come into the next video and I'll see you there we'll practice using the more flexible object

98
00:06:32,150 --> 00:06:35,690
orientated map plot named API directly with pandas.