1
00:00:00,390 --> 00:00:08,040
We have been mentioning about some models which are applicable on only time series data, which does

2
00:00:08,040 --> 00:00:16,800
not have trend and seasonality, but there is a special term for such cities such these are called stationary

3
00:00:16,800 --> 00:00:17,460
cities.

4
00:00:18,000 --> 00:00:26,370
Well, technically, a stationary time series is one whose statistical properties, such as mean variance,

5
00:00:26,790 --> 00:00:31,110
autocorrelation, etc., are all constant over time.

6
00:00:31,860 --> 00:00:38,640
For most business and economic activities, that Time series data is a fairly stationary.

7
00:00:39,120 --> 00:00:47,640
For example, if you look at the mean of the monthly sales of a company for 2010 and then for 2019,

8
00:00:48,090 --> 00:00:50,850
most probably the means will be different.

9
00:00:51,360 --> 00:00:57,330
That company would have either grown or declined in sales performance over 10.

10
00:00:57,990 --> 00:01:04,350
So since the mean is changing over time, this time series is not stationary.

11
00:01:04,440 --> 00:01:11,170
And we can not apply models such as we are or at Imar to forecast it.

12
00:01:11,910 --> 00:01:20,130
So to make a time series stationary, we just need to use different thing, as we have shown you earlier.

13
00:01:21,150 --> 00:01:26,000
We will use different thing to remove trend and seasonality book.

14
00:01:26,520 --> 00:01:32,730
And for most purposes, we will be able to use our models on the resulting series.

15
00:01:33,330 --> 00:01:41,940
And you can simply use models such as Satima or Sa'adi Max, which handles seasonality and trend.