1
00:00:00,900 --> 00:00:07,860
Here are the general's steps in Malta in the process of forecasting in the coming lectures.

2
00:00:08,160 --> 00:00:10,560
We will focus more on each of these steps.

3
00:00:11,130 --> 00:00:18,150
But in this lecture, we will just look at the overall process as an all data analysis.

4
00:00:18,540 --> 00:00:24,750
The process of forecasting begins with good definition, which means that we need to clearly set the

5
00:00:24,750 --> 00:00:29,040
objectives which we aim to achieve from the forecasting exercise.

6
00:00:30,750 --> 00:00:38,400
Then relevant data is collected and cleaned and explored using visualization tools.

7
00:00:39,840 --> 00:00:43,050
This is basically getting the data ready for analysis.

8
00:00:44,100 --> 00:00:45,570
And this is a big task.

9
00:00:46,590 --> 00:00:52,650
In fact, 50 to 60 percent of your time will be spent in getting the data ready for analysis.

10
00:00:54,910 --> 00:01:01,300
Once the details are ready, we select a set of prudential forecasting methods based on the nature of

11
00:01:01,300 --> 00:01:01,720
the data.

12
00:01:03,660 --> 00:01:10,020
All these different methods are applied and their performance is compared in terms of forecast accuracy

13
00:01:10,830 --> 00:01:17,460
and other measures related to what set good, the best method is then chosen.

14
00:01:18,150 --> 00:01:20,190
And it is used to generate forecast.

15
00:01:23,600 --> 00:01:33,170
Of course, the process does not end here because forecasting is typically an ongoing goal and forecast

16
00:01:33,260 --> 00:01:40,970
accuracy is monitored and sometimes the forecasting method is adapted or changed to accommodate the

17
00:01:40,970 --> 00:01:44,870
changes in the good or changes in the data over time.

18
00:01:47,720 --> 00:01:50,840
Also note the two set of arrows.

19
00:01:52,760 --> 00:01:56,240
These are indicating that part of the process are iterative.

20
00:01:57,290 --> 00:02:05,420
For instance, once the series is explored, one might determine that the series at hand cannot achieve

21
00:02:05,420 --> 00:02:10,020
the required goal leading to the collection of new or supplemented data.

22
00:02:11,900 --> 00:02:13,820
And that I dreaded process takes place here.

23
00:02:14,990 --> 00:02:21,860
When we are playing a forecasting method and evaluating its performance, the evaluation often leads

24
00:02:21,860 --> 00:02:28,970
to tweaking or adapting the existing matter or even trying out some new method.

25
00:02:30,830 --> 00:02:34,280
This is the process generally followed for forecasting.

26
00:02:35,240 --> 00:02:40,670
Keeping this process low in mind will help you approach the problem at hand in a structured way.

27
00:02:42,010 --> 00:02:45,030
Now, let's start discussing these tips in depth.

28
00:02:45,260 --> 00:02:45,830
One by one.
