0
1
00:00:00,620 --> 00:00:08,480
Seaborn's Web site looks like this. Seaborn is a Python module that's actually based on matplotlib
1

2
00:00:08,730 --> 00:00:15,970
but it has some nice extra features. You can think of seaborn like matplotlib on steroids.
2

3
00:00:16,230 --> 00:00:24,300
If matplotlib was Lance Armstrong then Seabourn is still Lance Armstrong.
3

4
00:00:24,450 --> 00:00:27,150
Lance Armstrong on the Tour de France.
4

5
00:00:27,900 --> 00:00:35,160
So let's import the Seaborn module at the top of the notebook as sns. I'm going to write
5

6
00:00:35,190 --> 00:00:45,060
"import seaborn as sns" and hit Shift+Enter. In the cell below the histogram we're gonna make use of one
6

7
00:00:45,060 --> 00:00:46,600
of seaborne functions.
7

8
00:00:46,830 --> 00:00:59,760
We're gonna make use of the distplot function, so "sns.distplot(data[
8

9
00:01:01,270 --> 00:01:13,280
'PRICE']) and then Enter, "plt.show()". Seaborn's distplot function
9

10
00:01:13,690 --> 00:01:21,320
only needs a single argument, namely the data it's meant to plot and we're gonna plot the target values and
10

11
00:01:21,410 --> 00:01:23,720
we're gonna see how this compares with
11

12
00:01:23,720 --> 00:01:27,990
matplotlib's histogram function. So let me hit Shift+Enter
12

13
00:01:28,820 --> 00:01:38,210
and here we can see that straight out of the box we get a very beautiful graph. We get both a histogram
13

14
00:01:38,840 --> 00:01:43,730
and an estimate of the probability density function.
14

15
00:01:44,120 --> 00:01:53,230
The probability density function is the darker squiggly line that's superimposed on top of the histogram.
15

16
00:01:53,300 --> 00:02:00,200
This is the line that estimates the distribution of the data and I think this is a neat little addition
16

17
00:02:00,260 --> 00:02:06,190
that Seaborn provides as an add on to the histogram.
17

18
00:02:06,230 --> 00:02:11,950
Now of course we can make the seaborn plot look a lot more like matplotlib.
18

19
00:02:12,140 --> 00:02:14,810
So for starters I could size it differently.
19

20
00:02:14,810 --> 00:02:18,580
So I'm going to take this line of code here.
20

21
00:02:18,650 --> 00:02:20,270
Copy it.
21

22
00:02:20,360 --> 00:02:22,640
Go down here.
22

23
00:02:22,640 --> 00:02:33,440
Paste it, press Shift+Enter can see that's a bit larger but easier to see. But we can also modify the
23

24
00:02:33,440 --> 00:02:34,510
number of arguments
24

25
00:02:34,520 --> 00:02:44,390
this distplot here takes. So it also has the bins argument that we can give it and I can give it 50.
25

26
00:02:45,770 --> 00:02:56,270
So now this ties out with what we see up here from matplotlib, and then we can also show and hide the two
26

27
00:02:56,300 --> 00:02:57,610
bits of the graph here.
27

28
00:02:57,680 --> 00:03:03,200
The PDF or probability density function and the histogram.
28

29
00:03:03,200 --> 00:03:14,790
So if I come up here to the arguments and I put "hist = False", with a capital F, and hit Shift+Enter
29

30
00:03:14,960 --> 00:03:22,610
it hides the histogram and we're just left with the probability density function which is the estimate
30

31
00:03:22,610 --> 00:03:25,700
for the distribution.
31

32
00:03:25,700 --> 00:03:27,470
And of course we can also do it the other way around.
32

33
00:03:27,470 --> 00:03:41,480
So if I write "kde = False" and then change the hist argument to True, then we're hiding the probability
33

34
00:03:41,480 --> 00:03:48,980
density function and just showing the histogram. And in terms of styling you can also grab a different
34

35
00:03:48,980 --> 00:03:49,690
color here.
35

36
00:03:50,130 --> 00:03:56,660
So if I want to go with a yellow palette for example pick this hex code, go back here,
36

37
00:03:58,380 --> 00:04:07,080
and then add the color argument as a keyword and paste the hex code in, hit Shift+Enter and then we'll see
37

38
00:04:07,080 --> 00:04:13,100
it like so. Now in retrospect I don't think this was actually a wise choice.
38

39
00:04:13,110 --> 00:04:15,330
This is very, very low contrast.
39

40
00:04:15,420 --> 00:04:23,730
I might have to go for something like this which is a bit more Amber and substitute that for the yellow.
40

41
00:04:24,930 --> 00:04:28,180
And this makes the histogram a lot more clear.
41

42
00:04:28,230 --> 00:04:33,230
All right so I think this is a really good time for a challenge.
42

43
00:04:33,300 --> 00:04:42,360
See if you can investigate and visualize two other features of the dataset and I also have a question
43

44
00:04:42,360 --> 00:04:43,350
for you.
44

45
00:04:43,440 --> 00:04:49,350
How many rooms do you think that the average property in Boston has?
45

46
00:04:49,410 --> 00:04:50,030
What do you think?
46

47
00:04:50,580 --> 00:04:54,210
What's the total number of rooms for a property in Boston?
47

48
00:04:54,210 --> 00:04:58,290
So including bedrooms, living rooms, bathrooms etc.
48

49
00:04:58,400 --> 00:05:00,050
Have you made your guess?
49

50
00:05:00,360 --> 00:05:04,280
Okay, so for this challenge we're gonna be checking the answer.
50

51
00:05:04,280 --> 00:05:07,580
We're gonna have a look at the RM feature.
51

52
00:05:07,730 --> 00:05:15,500
Can you write the Python code to plot this on a histogram and visualize the number of rooms per dwelling
52

53
00:05:16,310 --> 00:05:23,120
and then figure out what the average number of rooms is for a property in Boston Massachusetts?
53

54
00:05:23,120 --> 00:05:25,280
I think this would be fairly straightforward.
54

55
00:05:25,300 --> 00:05:31,240
Uh you can use either matplotlib or seaborn to generate your histogram but
55

56
00:05:31,430 --> 00:05:35,190
maybe have a think about what you would expect beforehand.
56

57
00:05:35,300 --> 00:05:36,080
Ready?
57

58
00:05:36,690 --> 00:05:38,710
Here's the solution.
58

59
00:05:39,480 --> 00:05:48,030
So what I'm gonna do is I'm going to take this cell here and I'm going to copy the cell, come down here
59

60
00:05:48,630 --> 00:05:56,870
and "Edit", paste the cell and then what 
I'm going to do is I'm going to change the code here, so I'm gonna 
60

61
00:05:56,970 --> 00:06:00,240
substitute RM for price.
61

62
00:06:00,240 --> 00:06:02,690
I'm going to delete the bins.
62

63
00:06:02,700 --> 00:06:06,590
I think this is no longer what we want.
63

64
00:06:06,600 --> 00:06:17,130
I'm also gonna pick a different color so maybe I'm gonna go for teal. It was dark teal here, just to differentiate
64

65
00:06:17,250 --> 00:06:25,350
the chart a little bit and then I'm going to change the x label here, because we're no longer looking
65

66
00:06:25,350 --> 00:06:27,550
at the price we're looking at
66

67
00:06:27,570 --> 00:06:38,910
then number of rooms. In fact this is the average number of rooms. And I'm going to hit Shift+Enter
67

68
00:06:41,700 --> 00:06:51,000
and the chart that I get back looks like this. We can see that the highest bar is around the number 6.
68

69
00:06:51,040 --> 00:06:57,640
In other words, most of the properties in Boston have around 6 rooms.
69

70
00:06:57,710 --> 00:07:04,000
I don't know about you but that actually seems pretty big actually, because if you think about it,
70

71
00:07:04,000 --> 00:07:06,700
bedroom, living room, kitchen, bathroom,
71

72
00:07:06,700 --> 00:07:09,910
that adds up to 4. If you add another bedroom,
72

73
00:07:09,910 --> 00:07:13,820
then you have 5, but the average is 6.
73

74
00:07:13,840 --> 00:07:20,740
So I don't know if they have a linen closet that they're counting or they have two bathrooms or three bedrooms.
74

75
00:07:20,740 --> 00:07:26,080
Now of course we don't know anything about the size of the rooms, but still I'm going to venture a guess
75

76
00:07:26,080 --> 00:07:30,810
that six rooms on average doesn't seem to be too shabby.
76

77
00:07:31,200 --> 00:07:34,330
I certainly wouldn't mind living somewhat with 6 rooms.
77

78
00:07:34,330 --> 00:07:38,280
That's palatial by London standards.
78

79
00:07:38,360 --> 00:07:44,320
Now we just eyeballed this graph to figure out the average number of rooms.
79

80
00:07:44,440 --> 00:07:50,880
What if we wanted to know an exact number for the average? Well pandas
80

81
00:07:50,890 --> 00:07:53,110
also has you covered for that.
81

82
00:07:53,560 --> 00:08:04,270
So what we can do is come down here and access our our RM data series, so "data['
82

83
00:08:04,360 --> 00:08:17,830
RM']", put a dot after it, and then call the mean() function. So "data['RM'].mean()".
83

84
00:08:18,470 --> 00:08:24,960
So here we're calling the mean method on our series object and hitting Shift Enter,
84

85
00:08:25,190 --> 00:08:29,630
we see that the true average is 6.28.