1
00:00:00,550 --> 00:00:01,800
Welcome back.

2
00:00:01,810 --> 00:00:05,950
In this section we're going to look at num pie.

3
00:00:06,070 --> 00:00:13,750
And now I'm very excited about this section because num pie is basically the backbone of all data science

4
00:00:13,750 --> 00:00:16,610
machine learning and numerical computing in Python.

5
00:00:16,840 --> 00:00:21,390
And that's actually what num pie stands for in a sense for numerical python.

6
00:00:21,610 --> 00:00:24,200
And what is num pie.

7
00:00:24,220 --> 00:00:29,430
Well in all of machine learning no matter what kind of data that you're working with.

8
00:00:29,530 --> 00:00:35,350
Now this could be a data frame like we saw in the previous section on panders num pi is going to form

9
00:00:35,350 --> 00:00:42,790
the foundation of turning your data into a series of numbers and then what a machine learning algorithm

10
00:00:42,790 --> 00:00:47,350
will do is work out the patterns in those numbers.

11
00:00:47,350 --> 00:00:54,250
So if we refer back to our machine learning modelling framework we've got Step 1 is Problem Definition

12
00:00:54,550 --> 00:00:59,940
Step 2 is Data Evaluation features modelling and experiments.

13
00:01:00,220 --> 00:01:06,340
And if we have a look at what tools we can use in this section we're going to be focused on num pi num

14
00:01:06,340 --> 00:01:08,950
pi comes into the data analysis section.

15
00:01:09,010 --> 00:01:14,800
However as you'll see in the future and what we just reference before is that machine learning although

16
00:01:14,800 --> 00:01:20,270
it's referred to as modelling is essentially taking data turning it into numbers.

17
00:01:20,350 --> 00:01:25,870
And in this case with NUM pi usually the form of a nun pi array which will have plenty of experience

18
00:01:25,870 --> 00:01:30,280
with after this section and then finding patterns in those numbers.

19
00:01:30,280 --> 00:01:40,340
So almost across this entire pipeline num PI will come into play now why num pi or the main reason we

20
00:01:40,340 --> 00:01:43,280
use num pi instead of pure Python.

21
00:01:43,280 --> 00:01:49,400
I mean you can do numerical calculations in Python but usually if we want to do large scale numerical

22
00:01:49,400 --> 00:01:54,450
calculations we refer to num PI because it's fast.

23
00:01:54,650 --> 00:02:02,110
And now the reason why it's fast is because behind the scenes there's optimizations written in C code.

24
00:02:02,240 --> 00:02:09,440
The beautiful thing about the NUM pi package is that you can use these C code optimizations using pure

25
00:02:09,440 --> 00:02:10,150
Python.

26
00:02:10,220 --> 00:02:16,580
That means you don't have to know C to be able to use num pi and Python it is a bit easier to write

27
00:02:16,580 --> 00:02:17,460
than pure C.

28
00:02:17,480 --> 00:02:20,230
So that's the advantage of using num pi.

29
00:02:20,300 --> 00:02:27,170
Now it gets these optimizations in speed through a process called vector ization via a term num pi uses

30
00:02:27,590 --> 00:02:34,610
called broadcasting which is essentially just avoiding writing loops when you're writing calculations

31
00:02:35,000 --> 00:02:40,940
because if you're writing loops to make calculations such as in Python to calculate over a list that

32
00:02:40,940 --> 00:02:46,640
can form some eventual slowdowns if you have a list say of of millions of different numbers which becomes

33
00:02:46,700 --> 00:02:52,340
common practice when we're working on machine learning projects and as we mentioned before num pi is

34
00:02:52,340 --> 00:02:58,790
the backbone of other Python scientific packages now pi actually forms the basis of the pandas package

35
00:03:00,290 --> 00:03:02,510
so what are we going to cover in this section.

36
00:03:03,050 --> 00:03:07,880
Well of course we're always going to be focused on the most useful functions the ones that you can start

37
00:03:07,880 --> 00:03:09,570
using straight away.

38
00:03:09,590 --> 00:03:14,150
We'll have a look at different nameplate data types and attributes of those data types.

39
00:03:14,150 --> 00:03:19,310
Now the main one is an NDA Ray but we'll have a look at that in another video.

40
00:03:19,400 --> 00:03:21,910
We'll look at how to create arrays are raised.

41
00:03:21,920 --> 00:03:24,370
You can think of as a list of numbers.

42
00:03:24,470 --> 00:03:27,690
We'll have a look at how to view arrays and matrices.

43
00:03:27,710 --> 00:03:34,190
Now matrix or a matrices you can imagine as a combination of different arrays will then look at how

44
00:03:34,190 --> 00:03:40,730
to manipulate and compare arrays will then finish up with sorting arrays which is sorting out different

45
00:03:40,730 --> 00:03:47,540
lists of numbers and then some practical use cases of where arrays get used or where num pi can be used

46
00:03:47,540 --> 00:03:55,090
to turn data into lists of numbers so where can you get help if you get stuck.

47
00:03:55,110 --> 00:04:00,510
Well the steps you want to try and follow is to follow along with the code so as we're going through

48
00:04:00,510 --> 00:04:06,390
the video as we're working through our Jupiter notebook through the introduction to num pi lessons you

49
00:04:06,390 --> 00:04:08,280
want to follow along as best you can.

50
00:04:08,310 --> 00:04:12,000
Now if you can't keep pace or if I'm going too fast that's all right.

51
00:04:12,000 --> 00:04:13,220
Feel free to stop it.

52
00:04:13,290 --> 00:04:18,360
Go back practice something on your own but you should be practicing writing the code and running it

53
00:04:18,390 --> 00:04:19,980
as much as possible.

54
00:04:20,250 --> 00:04:21,960
So that's step to try it for yourself.

55
00:04:21,960 --> 00:04:23,150
Try to run the code.

56
00:04:23,280 --> 00:04:28,980
Remember the saying If in doubt run the code if you get stuck on a certain function or a certain thing

57
00:04:28,980 --> 00:04:29,780
that we've covered.

58
00:04:29,910 --> 00:04:35,520
You can try to search for it here you'll probably come across stack overflow or the official name pi

59
00:04:35,520 --> 00:04:37,380
documentation.

60
00:04:37,380 --> 00:04:39,350
Then after that once you've searched for.

61
00:04:39,660 --> 00:04:43,230
If you find something on stack overflow or the NUM pi documentation.

62
00:04:43,230 --> 00:04:48,720
The important thing is to keep trying again enter the code that you find in the Jupiter notebook.

63
00:04:48,840 --> 00:04:50,040
Practice running it.

64
00:04:50,040 --> 00:04:55,890
If in doubt run the code and then finally if you're still stuck don't forget you can always ask for

65
00:04:55,890 --> 00:04:56,330
help.

66
00:04:56,340 --> 00:05:01,050
There's no such thing as a stupid question because if you have a question it's likely someone else has

67
00:05:01,050 --> 00:05:02,230
the same question.

68
00:05:02,430 --> 00:05:06,810
You can ask on stack overflow or in any other resource you can find.

69
00:05:06,900 --> 00:05:10,480
All right with that being said that's an introduction of num PI.

70
00:05:10,590 --> 00:05:14,610
It's the foundation it's the backbone of other data science and scientific packages.

71
00:05:14,610 --> 00:05:16,020
But let's get in in code.

72
00:05:16,050 --> 00:05:17,090
Let's see what it's all about.