1
00:00:07,080 --> 00:00:11,250
In this section, we will cover the most important part of data analysis.

2
00:00:12,230 --> 00:00:14,270
This is called data pre processing.

3
00:00:15,760 --> 00:00:21,910
As I told you in the introduction to Machine Learning, Section four, practicing data analysts, nearly

4
00:00:21,940 --> 00:00:25,300
80 percent of their time is spent in data processing.

5
00:00:26,410 --> 00:00:31,390
And how well your model will perform eventually depends on how well you have prepared your data.

6
00:00:33,180 --> 00:00:37,110
So let's start with the first thing you should know when you are trying to solve a problem.

7
00:00:38,240 --> 00:00:41,060
You should know about the business context of that problem.

8
00:00:41,630 --> 00:00:47,810
You should know what all factors are important and have impact on the variables of interest.

9
00:00:49,480 --> 00:00:55,720
Only if we know the business, we will be able to identify relevant variables and gather required data

10
00:00:55,810 --> 00:00:56,860
for the analysis.

11
00:00:58,290 --> 00:00:59,070
Remember this.

12
00:00:59,540 --> 00:01:03,090
The quality of your input will decide quality of your output.

13
00:01:04,180 --> 00:01:07,840
Now, there are two primary ways to gather information about business.

14
00:01:08,710 --> 00:01:14,830
One is to gather it yourself by talking to relevant people or stakeholders of that business.

15
00:01:15,460 --> 00:01:22,210
People who are impacted by the problem at hand, or you can also gather information by actually doing

16
00:01:22,210 --> 00:01:23,320
the things yourself.

17
00:01:24,160 --> 00:01:28,780
For example, if you have to increase sales of a particular product.

18
00:01:29,960 --> 00:01:32,260
Tried to go and sell it yourself in the market.

19
00:01:33,490 --> 00:01:36,950
Go and try to buy it, also to understand customers perspective.

20
00:01:38,500 --> 00:01:43,390
Gathering information by yourself is called doing primary research.

21
00:01:44,760 --> 00:01:50,480
In secondary research, you read or listen to the information gathered by others.

22
00:01:51,410 --> 00:01:58,310
Maybe there is some industry related research report or some study done by some consulting firm or someone

23
00:01:58,310 --> 00:02:01,050
else has worked on the same problem before.

24
00:02:02,360 --> 00:02:06,710
We should look at methodology and findings of those earlier works.

25
00:02:06,800 --> 00:02:07,250
Also.

26
00:02:09,910 --> 00:02:10,870
That is an example.

27
00:02:12,060 --> 00:02:15,880
Suppose you're part of a company which is online.

28
00:02:16,530 --> 00:02:19,650
And the problem you are facing is off guard abandonment.

29
00:02:19,920 --> 00:02:26,130
That is a lot of your customers add product to their card, but do not actually putties it.

30
00:02:27,850 --> 00:02:30,940
This is a fairly common problem for online businesses.

31
00:02:33,410 --> 00:02:37,990
Now, if you are given the task to tackle this problem, where do you start?

32
00:02:40,070 --> 00:02:42,900
First, we should go and talk with the relevant teams involved.

33
00:02:45,050 --> 00:02:51,050
Usually there is a marketing team and a product team which cross-functional handle the online sales

34
00:02:51,050 --> 00:02:51,560
process.

35
00:02:52,400 --> 00:02:53,650
We should go and talk to them.

36
00:02:55,800 --> 00:03:00,810
Then we should try choosing from this site ourselves also to understand what is the customer journey

37
00:03:00,810 --> 00:03:01,320
online.

38
00:03:02,730 --> 00:03:04,440
All this will be primarily search.

39
00:03:05,530 --> 00:03:11,350
As part of the cavity search, we should look for previous studies on gut abandonment within and outside

40
00:03:11,350 --> 00:03:12,130
the organization.

41
00:03:14,440 --> 00:03:20,230
Once we have done all this, we would have a fairly good idea of what all variables and what type of

42
00:03:20,230 --> 00:03:22,060
model we are going to use.

43
00:03:23,670 --> 00:03:25,830
Next task is to get Dee dee dum.