1
00:00:01,020 --> 00:00:04,980
Now that our data is ready, we can start building a classification model.

2
00:00:07,080 --> 00:00:14,580
There are many classification techniques or classifiers available which can be used to predict a qualitative

3
00:00:15,000 --> 00:00:17,280
or categorical type of responsible evil.

4
00:00:18,570 --> 00:00:23,250
Here's a list of all the classification techniques we are going to learn in the coming set of lectures.

5
00:00:25,350 --> 00:00:31,800
These three techniques are not very computationally heavy on the system, and they give fairly good

6
00:00:31,860 --> 00:00:33,270
accuracy in predictions.

7
00:00:35,730 --> 00:00:39,000
What we are going to do now is we will pick up one technique.

8
00:00:39,960 --> 00:00:41,790
Understand the intuition behind it.

9
00:00:41,790 --> 00:00:42,930
A few short lectures.

10
00:00:44,450 --> 00:00:46,590
And then we will learn how to implement it.

11
00:00:46,820 --> 00:00:47,930
And this software package.

12
00:00:51,240 --> 00:00:58,800
Lastly, we will also discuss the output and the quality of the model that we have created in this process.

13
00:00:59,160 --> 00:01:02,630
I will alter they knew when to prefer and when north to preffered.

14
00:01:02,790 --> 00:01:03,840
Each of these techniques.

15
00:01:06,510 --> 00:01:12,450
For practice purposes, you have one data set on which you must be working side by side after every

16
00:01:12,450 --> 00:01:12,900
two related.

17
00:01:14,550 --> 00:01:15,210
By now.

18
00:01:15,450 --> 00:01:18,900
That would also be preprocessed and ready for further analysis.

19
00:01:20,850 --> 00:01:23,490
I will share with you one more data set at the end.

20
00:01:24,480 --> 00:01:30,750
Once we have discussed all these techniques, the point of that data set is that you should test your

21
00:01:30,750 --> 00:01:35,850
intuition of which classification method is best suited for which problem.

22
00:01:37,380 --> 00:01:40,550
So just take that data set and figure out the best predictive model for it.

23
00:01:44,350 --> 00:01:50,230
So before we close this lecture and start with these techniques, let us put down the two title business

24
00:01:50,230 --> 00:01:53,480
questions that you may want to answer from.

25
00:01:53,770 --> 00:01:54,640
Building this model.

26
00:01:57,720 --> 00:01:59,580
The first US prediction question.

27
00:02:00,840 --> 00:02:05,790
Prediction question is will the House be sold within three months of getting listed?

28
00:02:07,320 --> 00:02:16,020
So if you remember, we had the data of houses, the asking price and some of the attributes and the

29
00:02:16,020 --> 00:02:22,080
response variable for which we want to predict the value is whether it is whether it will be sold in

30
00:02:22,140 --> 00:02:23,010
three months or not.

31
00:02:24,480 --> 00:02:26,100
So that is a prediction question.

32
00:02:27,730 --> 00:02:32,750
And the prediction question, we do not worry about the effect of individual variables on the output.

33
00:02:33,350 --> 00:02:38,480
Our aim is to predict as accurately as possible the value of the response variable.

34
00:02:41,020 --> 00:02:43,750
The second type of question is inferential question.

35
00:02:45,670 --> 00:02:50,620
Here we are trying to infer the importance and impact of each independent variables.

36
00:02:51,010 --> 00:02:52,510
So these all variables.

37
00:02:53,440 --> 00:02:57,190
What is the impact of each of these variables on the response variables?

38
00:02:59,510 --> 00:03:06,500
So the inferential question here is how accurately I can estimate the effect of each of these variables

39
00:03:06,650 --> 00:03:14,810
on the response of even so, we will try to find the answers to these two questions using all our different

40
00:03:14,870 --> 00:03:15,680
classifiers.
