1
00:00:00,270 --> 00:00:06,530
In this lecture, we will look at different Bhutto's relation techniques that are available for war

2
00:00:06,570 --> 00:00:07,710
time series data.

3
00:00:09,990 --> 00:00:15,120
First, we will start with floating line chad for our data cities.

4
00:00:16,560 --> 00:00:21,240
Then we will look at how to zoom in at any particular window of time.

5
00:00:22,320 --> 00:00:28,410
Then we will look at how to a trend line for a scatter plot of our data cities.

6
00:00:30,090 --> 00:00:38,040
And then we will see how to use release just Air-Crane off of it daytime better cities by removing seasonality.

7
00:00:39,480 --> 00:00:43,680
And at last, we will look at lotting like plots.

8
00:00:45,010 --> 00:00:49,110
Plotting autocorrelation plots, four legged data.

9
00:00:53,070 --> 00:00:56,310
So let's first start with plotting line, Chuck.

10
00:00:57,940 --> 00:00:59,430
We will import my plot.

11
00:01:01,890 --> 00:01:05,790
We will be using same data that we were using earlier.

12
00:01:07,230 --> 00:01:15,050
And creating another data frame that is due to the relation underscore D.F. on let's look at first and

13
00:01:15,060 --> 00:01:16,410
values of this dataset.

14
00:01:17,880 --> 00:01:23,440
So, as you know, the first column is off dates and the second column is off.

15
00:01:23,600 --> 00:01:23,850
No.

16
00:01:24,130 --> 00:01:33,330
But notice here that indexes are ranging from zero one to three and so on.

17
00:01:35,610 --> 00:01:42,180
So to plot a line, Chuck, you just have to mention the values that you want on your y axis.

18
00:01:42,480 --> 00:01:44,610
So here we want Berts.

19
00:01:45,300 --> 00:01:48,120
So we are just calling database underscore D.F..

20
00:01:48,570 --> 00:01:50,370
Then we are selecting work column.

21
00:01:50,640 --> 00:01:53,220
And then we are using dot plot method.

22
00:01:53,670 --> 00:01:55,590
So if you just write dot plot matter.

23
00:01:55,860 --> 00:02:00,240
It will plot that particular value as a line plot.

24
00:02:00,550 --> 00:02:01,740
So if you run this,

25
00:02:05,300 --> 00:02:08,760
this is the plot we are getting on the Y axis.

26
00:02:09,030 --> 00:02:11,280
We have the birds value.

27
00:02:11,700 --> 00:02:12,510
You can see.

28
00:02:13,080 --> 00:02:18,120
You can see that trend then variation in number of birds using this plot.

29
00:02:18,690 --> 00:02:26,040
And on the x axis, you can see we do not have the date, but we have the index value.

30
00:02:26,790 --> 00:02:33,810
So this first value is the value of bird that is present at zero to index.

31
00:02:35,730 --> 00:02:36,620
Which is 35.

32
00:02:36,960 --> 00:02:41,040
So this plot is starting from 35 and so on.

33
00:02:41,370 --> 00:02:51,210
So on the x axis, we have indexes value, but it will be better to have added value instead of this

34
00:02:51,330 --> 00:02:55,440
index value since index value does not make any sense.

35
00:02:56,910 --> 00:03:05,280
So to get the data value, we will first change our index value to over time values and then we will

36
00:03:05,280 --> 00:03:06,510
plot the same plot.

37
00:03:07,530 --> 00:03:14,580
So here I am changing the indexes of my database D.F. to database date column.

38
00:03:15,000 --> 00:03:17,580
So let's run this.

39
00:03:17,880 --> 00:03:22,570
I'm calling database DOT Index equal to database underscore D.F..

40
00:03:23,070 --> 00:03:24,870
Then we are selecting that data column.

41
00:03:25,530 --> 00:03:32,130
This command will change the value of indexes from zero and so on to this data values.

42
00:03:33,380 --> 00:03:34,320
NetSuite on this

43
00:03:36,950 --> 00:03:38,720
to again look at 30 duffing.

44
00:03:39,030 --> 00:03:40,830
I will rerun this.

45
00:03:41,490 --> 00:03:41,910
Come on.

46
00:03:43,260 --> 00:03:45,510
Notice here we have indexes from zero.

47
00:03:46,220 --> 00:03:52,380
If I run this again now, we have built in place of index.

48
00:03:52,920 --> 00:03:58,790
So my new index are in the form of data values and they are not in the form of zero.

49
00:03:58,800 --> 00:03:59,970
One, two, one.

50
00:04:00,750 --> 00:04:02,280
And the other numerical value.

51
00:04:03,150 --> 00:04:11,010
So if I again plot the same plot using database and as for the book and then calling the method dot

52
00:04:11,010 --> 00:04:11,460
plot,

53
00:04:15,360 --> 00:04:16,050
you can see.

54
00:04:16,680 --> 00:04:21,780
Now I have bid values instead of this index value on my x axis.

55
00:04:22,530 --> 00:04:27,900
So that so you plot a simple date time series and do a line plot.

56
00:04:28,920 --> 00:04:33,810
Now let's look at how to zoom in at any particular interval of time.

57
00:04:34,770 --> 00:04:41,040
It is very difficult for me to understand the pattern between first gen two today.

58
00:04:41,040 --> 00:04:41,630
First Gen.

59
00:04:42,240 --> 00:04:43,770
Since the values are clammed up.

60
00:04:44,640 --> 00:04:50,170
So what I can do is I can just plot values between four subject to today.

61
00:04:50,230 --> 00:04:50,880
First off.

62
00:04:51,160 --> 00:04:51,470
Gen.

63
00:04:52,440 --> 00:04:58,840
And this will help me in focusing at a particular time interval instead of looking at the overall time

64
00:04:58,840 --> 00:04:59,220
frame.

65
00:05:01,470 --> 00:05:04,700
So suppose I want to get the data.

66
00:05:07,610 --> 00:05:17,300
From dates with Queen Foxtrot June to 10th of June, so I can just cleared another date frame by selecting

67
00:05:17,480 --> 00:05:19,310
a segment of my original date.

68
00:05:19,970 --> 00:05:23,600
So I am creating database and this could be a two.

69
00:05:24,290 --> 00:05:33,880
And then I am selecting values where date lies between first of June 1959 to of 10, 1959.

70
00:05:35,320 --> 00:05:41,030
We have already seen this how to select a particular enteral earlier in data preprocessing.

71
00:05:43,580 --> 00:05:54,960
So let's find this and let's look at word database underscore B to do different that we have to do just.

72
00:06:00,000 --> 00:06:08,400
You can see we have values from second of Jan since we selected greater than sine, not greater than

73
00:06:08,400 --> 00:06:08,970
equal to St..

74
00:06:10,290 --> 00:06:14,190
So we have values from second of 10 to 10th of June.

75
00:06:15,300 --> 00:06:21,140
Now, to load this values, I can just call dot plot method.

76
00:06:24,600 --> 00:06:25,660
You can see this is the line.

77
00:06:25,800 --> 00:06:30,210
Lord, I'm getting just look at the difference here.

78
00:06:30,270 --> 00:06:34,200
All these values are between these two points.

79
00:06:34,380 --> 00:06:39,180
So it's really difficult for me to get any insight from this line to.

80
00:06:39,630 --> 00:06:46,550
Whereas if I zoom in at any particular interval, I can look at and make some insights from this lecture.

81
00:06:49,710 --> 00:06:53,740
The next thing we discuss in our theory lecture is trend line

82
00:06:56,480 --> 00:06:57,500
here for line.

83
00:06:57,810 --> 00:07:04,560
We are going to use Seabourne Library since it is very easy, from Seabourne Library to Blätter strangling.

84
00:07:08,360 --> 00:07:15,320
You can do it with Matt Plourde as well, but it will be a lengthy process and it is better to use Seabourne

85
00:07:16,160 --> 00:07:19,670
since it is very easy to plot Trainline using Seabourne.

86
00:07:21,230 --> 00:07:25,430
So let's import Seabourne now in Seabourne.

87
00:07:25,550 --> 00:07:31,910
We have a special plot known as Rect Plot, which will automatically Blätter tramline for you.

88
00:07:34,490 --> 00:07:37,970
So here we will be calling Rect Plot.

89
00:07:38,690 --> 00:07:42,320
And here we need to give our X and Y values.

90
00:07:44,240 --> 00:07:45,590
So four X values.

91
00:07:45,790 --> 00:07:53,940
I'm giving my index values and for Y value I'm giving the best values of forward.

92
00:07:53,990 --> 00:07:55,230
Be up to date frame.

93
00:07:57,080 --> 00:07:59,360
So let's look at this.

94
00:08:02,360 --> 00:08:12,290
You can see we have a scatterplot where we have indexers values on the x axis and birth values on the

95
00:08:12,290 --> 00:08:13,040
Y axis.

96
00:08:14,420 --> 00:08:18,950
Also, notice that here we are using index values instead of date values.

97
00:08:20,680 --> 00:08:24,640
Since for a scatterplot, we need numeric data.

98
00:08:25,360 --> 00:08:31,720
So if I just run that plot with our date data, you will get an error.

99
00:08:32,290 --> 00:08:39,700
And we were not able to load this graph just because on the x axis, we have some other data other than

100
00:08:39,760 --> 00:08:40,600
numeric data.

101
00:08:41,320 --> 00:08:44,020
So for scatterplot, you need numeric data.

102
00:08:44,350 --> 00:08:46,660
And that's why we're using it next values.

103
00:08:48,400 --> 00:08:52,600
So, again, you can see that there is a line plotted in between.

104
00:08:52,930 --> 00:08:55,720
This is the regression line or a trend line.

105
00:08:56,380 --> 00:08:58,900
You can see that is an increasing trend.

106
00:08:59,710 --> 00:09:04,300
So using trend line, you can easily identified trend in your data.

107
00:09:04,900 --> 00:09:14,020
So if the slope is positive here, you can see with the increase in values of X, my trend line values

108
00:09:14,110 --> 00:09:15,070
is going up.

109
00:09:15,430 --> 00:09:23,080
So there is a positive slope and you can say that there is a positive trend in your data.

110
00:09:24,970 --> 00:09:30,880
So in any case, if you want to identify whether your data have trend or not, just plot direct plot

111
00:09:31,090 --> 00:09:33,940
and see if the slope of the line is positive.

112
00:09:34,180 --> 00:09:35,200
That is some trend.

113
00:09:35,560 --> 00:09:39,010
And if the line is flat, then there is no trend.

114
00:09:41,050 --> 00:09:42,830
No on.

115
00:09:43,150 --> 00:09:44,440
This is a linear trend line.

116
00:09:44,680 --> 00:09:48,090
You can also plot polynomial or quadratic trendline.

117
00:09:48,640 --> 00:09:57,400
So if you want to have a quadratic trainline, that is the equation of your trend line as Vike to A

118
00:09:57,520 --> 00:10:00,470
X squared plus B, X plus C.

119
00:10:00,910 --> 00:10:09,190
So if you want any such trend line, you can provide another parameter here, which is the order parameter.

120
00:10:09,580 --> 00:10:11,440
By default, the order value is one.

121
00:10:11,560 --> 00:10:20,500
So that's why when we just direct plot with X and Y values, the order was one and the default trend

122
00:10:20,500 --> 00:10:21,850
line was a linear trend line.

123
00:10:22,400 --> 00:10:29,830
But if you provide the order parameter here using the same that plot method, a fibroid already equal

124
00:10:29,830 --> 00:10:35,440
to this will lord the four predict trendline a fibroid already equal to three.

125
00:10:35,800 --> 00:10:37,820
This will Lawder Kubic trend line.

126
00:10:38,650 --> 00:10:45,220
If I provide all that equate to four or five, go to the trend line of order for or order five.

127
00:10:46,630 --> 00:10:49,480
So let's run this again with the quadratic.

128
00:10:52,990 --> 00:11:00,160
Can see we have the same graph, but this trend line here is not a straight line anymore.

129
00:11:00,640 --> 00:11:01,720
So Quadri decline.

130
00:11:02,560 --> 00:11:08,090
And you can also plot Kubic trend line or trend line with order four or five as well.

131
00:11:13,450 --> 00:11:20,920
Now, for the next two days, relation examples, we will be using another dataset that is monthly aircraft,

132
00:11:20,920 --> 00:11:22,380
mind blown dataset.

133
00:11:23,170 --> 00:11:32,020
So this is the dataset which contains data of how many miles that U.S. airline aircrafts are flying.

134
00:11:33,580 --> 00:11:38,120
Let's load this data set in two miles, underscore B if.

135
00:11:40,230 --> 00:11:50,810
Let's take a look at top five values of this data so you can see on the first column V of the month

136
00:11:50,860 --> 00:11:56,500
data where you have data from January 1963.

137
00:11:57,340 --> 00:12:00,550
And the data is at the interval of one month.

138
00:12:00,850 --> 00:12:05,770
So in the next call, I have data Fab 1963 and so on.

139
00:12:06,940 --> 00:12:13,030
And I have another column in which we have values of miles and millions.

140
00:12:13,810 --> 00:12:21,910
So in Jan, 1963, that total miles covered are six thousand eight hundred and twenty seven million

141
00:12:21,910 --> 00:12:22,480
miles.

142
00:12:23,080 --> 00:12:29,190
So I have this data, that first floor flautre, simple line chart.

143
00:12:34,240 --> 00:12:43,440
So this is the data, you can see we have a positive trend or what all the number of miles are increasing.

144
00:12:43,740 --> 00:12:47,070
And we also have a seasonal pattern in our data.

145
00:12:48,180 --> 00:12:53,150
So you can see ups and down at the constant interval of time.

146
00:12:53,760 --> 00:12:58,440
That is the indicator that our data have some seasonality.

147
00:13:01,540 --> 00:13:10,160
In legs lawter crime line on this, Fruita will be using as an as daughter plot here, our Axwell Leuser

148
00:13:10,240 --> 00:13:18,210
miles hundreds could be, if not indexed, taught values and our way values are just the minus number.

149
00:13:19,790 --> 00:13:27,050
So if you plot this, we have scatterplot and see the ups and downs in the scatterplot as well.

150
00:13:27,980 --> 00:13:30,910
And you can also see the trend line.

151
00:13:31,520 --> 00:13:38,450
The slope of this trend line is positive, which indicates that the number of miles are increasing over

152
00:13:38,510 --> 00:13:38,810
time.

153
00:13:40,250 --> 00:13:44,720
Next, we are going to discuss how to remove seasonality from our data.

154
00:13:45,740 --> 00:13:49,010
So in this data, you can see that we have a clear seasonal pattern.

155
00:13:52,940 --> 00:14:00,290
The method we discuss in order to re-elect her was to aggregate values at yearly level to remove the

156
00:14:00,290 --> 00:14:02,480
effect of monthly seasonality.

157
00:14:04,860 --> 00:14:12,210
So what we are trying to do here is we are aggregating our data at the year level.

158
00:14:12,810 --> 00:14:19,070
So I will be taking the average value of miles in 1963.

159
00:14:19,740 --> 00:14:24,090
Then I will take the average value of miles in 1964.

160
00:14:24,900 --> 00:14:28,470
And since a complete year is covering all the seasons.

161
00:14:29,310 --> 00:14:37,020
So finally, if I applaud the aggregate values of this, Miles, on each level, my seasonality will

162
00:14:37,020 --> 00:14:37,430
be gone.

163
00:14:39,980 --> 00:14:44,340
So first, let's create another column in our data.

164
00:14:45,310 --> 00:14:46,580
We are calling that column.

165
00:14:46,670 --> 00:14:47,130
Yes.

166
00:14:47,900 --> 00:14:51,490
And we are storing the easier information of the day.

167
00:14:52,880 --> 00:14:56,600
So in the first column, which is the month column, we have the complete date.

168
00:14:56,870 --> 00:14:58,310
Now we want another column.

169
00:14:58,390 --> 00:15:01,360
Years where we are just picking the e're value.

170
00:15:01,850 --> 00:15:09,900
So for all these values, I won the year column to take 1963 as the year value for all this for.

171
00:15:10,410 --> 00:15:11,990
For all these five reports.

172
00:15:12,560 --> 00:15:14,210
And of the year is 1964.

173
00:15:14,460 --> 00:15:16,940
It should take 1964 as the year.

174
00:15:18,190 --> 00:15:20,720
So we have already discussed how to do this.

175
00:15:21,430 --> 00:15:30,440
And our data prepossessing, which again will be using beauty and beauty sense for dating and dot here.

176
00:15:32,180 --> 00:15:33,680
So if I fed on this.

177
00:15:35,800 --> 00:15:37,350
Let's look at our data frame again.

178
00:15:40,020 --> 00:15:45,700
So now we have another problem here, which is taking the air value of this big family.

179
00:15:48,750 --> 00:15:51,840
Now, I want to aggregate data on this each level.

180
00:15:52,050 --> 00:15:59,370
So for all the records where it is 1963, I want to take the mean of all those values for for all the

181
00:15:59,370 --> 00:16:01,740
records where it is 1964.

182
00:16:02,250 --> 00:16:07,830
I want to take average or mean of my miles for those 12 records.

183
00:16:09,510 --> 00:16:10,540
So how to do that?

184
00:16:10,740 --> 00:16:13,290
We can do that using group by function.

185
00:16:15,360 --> 00:16:19,980
So we are first grouping our data on the basis of years.

186
00:16:20,610 --> 00:16:28,950
Then we are taking miles well use for all those loops and we are taking mean of those values.

187
00:16:29,100 --> 00:16:32,340
So for 1963, I have 12 values on Miles.

188
00:16:34,470 --> 00:16:41,700
Since there are four elements I can take, mean I can take some I can take maximum value.

189
00:16:41,730 --> 00:16:43,050
I can take minimum value.

190
00:16:43,170 --> 00:16:50,370
But here I am taking the mean value of this well values, then I will take another mean value of the

191
00:16:50,370 --> 00:16:53,640
next 12 values of 1964 and so on.

192
00:16:54,930 --> 00:16:56,460
So let's run this.

193
00:16:58,340 --> 00:16:59,470
Those in Texas.

194
00:16:59,490 --> 00:17:00,210
Very simple.

195
00:17:00,240 --> 00:17:03,390
First you have to mention the group by clause.

196
00:17:03,630 --> 00:17:12,450
So here we want to group our data using your data and then we want to mention what values are we going

197
00:17:12,450 --> 00:17:13,260
to get.

198
00:17:13,800 --> 00:17:20,250
So we want to be minus values and we want to take the mean of all the mind values.

199
00:17:21,330 --> 00:17:22,360
Let's run this.

200
00:17:24,640 --> 00:17:27,560
Now you can see that we have our data.

201
00:17:27,780 --> 00:17:30,990
We have the year value and the mean value of that year.

202
00:17:34,360 --> 00:17:36,870
So plotting this data is very easy.

203
00:17:36,960 --> 00:17:38,780
We will again use dot plot metric.

204
00:17:38,940 --> 00:17:40,200
So we have this data.

205
00:17:40,440 --> 00:17:44,630
I can just use dot plot in front of this dataset.

206
00:17:45,420 --> 00:17:53,040
You can also save this data and do another data frame and use that data frame and dot plot method plotted.

207
00:17:53,820 --> 00:17:56,100
But it's the same.

208
00:17:57,010 --> 00:18:00,600
I'm just using this.

209
00:18:01,470 --> 00:18:06,930
You can see that this is our data from the same series that we have plotted earlier.

210
00:18:07,110 --> 00:18:09,570
So earlier it was looking like this.

211
00:18:10,530 --> 00:18:14,280
This ups and downs are due to seasonal pattern in our data.

212
00:18:15,210 --> 00:18:21,150
Now, if I am aggregating all the values at the year level, I am reviewing seasonality.

213
00:18:21,480 --> 00:18:27,510
And now I'm just getting these analyzed chart for our data.

214
00:18:30,830 --> 00:18:34,460
The next Rusal election, my third is beating the leg block.

215
00:18:36,990 --> 00:18:42,500
So as discussed in the two re-elected leg, values are very important in forecasting.

216
00:18:43,760 --> 00:18:45,810
So mostly in forecasting problem.

217
00:18:46,350 --> 00:18:51,300
The value we want to forecast is highly dependent on the previous values.

218
00:18:52,800 --> 00:19:00,100
So suppose if you have a temperature data and you want to predict the temperature for tomorrow, it

219
00:19:00,100 --> 00:19:07,230
is more likely to be plus or minus three or four degrees as compared to two days temperature.

220
00:19:11,040 --> 00:19:15,520
So we have also discussed lagging value and our data people siccing.

221
00:19:18,480 --> 00:19:22,530
So let's first create a lag variable.

222
00:19:23,010 --> 00:19:24,540
We are calling a lag one.

223
00:19:25,500 --> 00:19:30,120
And here we are creating a lag variable for just one time period.

224
00:19:30,420 --> 00:19:39,060
So I can write minus underscore B minus M-m that this is the value we want to create like variables

225
00:19:39,060 --> 00:19:39,340
for.

226
00:19:40,020 --> 00:19:45,600
Then I can write, then I can call Lordship Shift method with argument as one.

227
00:19:46,560 --> 00:19:49,560
So this will shift all the values my one time.

228
00:19:51,330 --> 00:19:52,120
Let's run this.

229
00:19:52,140 --> 00:19:53,120
Let's look at her.

230
00:19:54,870 --> 00:19:55,710
So you can see.

231
00:19:55,830 --> 00:19:59,910
Now I have another call them, which is leg one column.

232
00:20:00,360 --> 00:20:09,210
And the values here are at the leg of one time period from minus M column.

233
00:20:10,600 --> 00:20:17,760
So now I have these two columns for my lagged variables and for minus Amum values.

234
00:20:18,570 --> 00:20:22,450
I can just create a scatterplot using these two values.

235
00:20:22,860 --> 00:20:25,390
My ex here is the leg value.

236
00:20:25,830 --> 00:20:28,530
And my way is the original value.

237
00:20:34,410 --> 00:20:40,140
So you can see this is the scatter plot of the leg, the value and the original value.

238
00:20:42,360 --> 00:20:47,550
And one thing you can notice is that there is some positive correlation between the two values.

239
00:20:48,540 --> 00:20:51,360
So you can see that if my leg value is high.

240
00:20:52,230 --> 00:20:53,900
My minus will lose also high.

241
00:20:54,960 --> 00:20:57,860
And if my leg value is low, Miles.

242
00:20:57,990 --> 00:20:59,040
Value is also low.

243
00:21:00,330 --> 00:21:05,700
This means that there is a positive correlation between the leg, the values of minus.

244
00:21:08,630 --> 00:21:15,470
Later on in our course, we will see why this leg values are important and why it is important to find

245
00:21:15,470 --> 00:21:18,410
out any correlation between the leg variables.

246
00:21:21,810 --> 00:21:29,910
Now, in this method of creating like Block B first created another column and then we plotted scatterplot,

247
00:21:30,580 --> 00:21:34,170
that is another simple method of plotting a Lego block.

248
00:21:35,700 --> 00:21:46,710
We can just import Lego block from Ponder's DOT plotting and we can directly call this method to load

249
00:21:46,710 --> 00:21:48,690
the Lego block by default.

250
00:21:48,720 --> 00:21:51,830
This will plot for Legoff one.

251
00:21:52,500 --> 00:21:56,760
So here I am not giving any value to shift the volume.

252
00:21:57,060 --> 00:21:59,940
This will automatically, by default, flawed.

253
00:22:00,120 --> 00:22:03,710
This variable with its leg off one time embitter.

254
00:22:05,220 --> 00:22:06,120
Let's run this.

255
00:22:07,350 --> 00:22:11,500
You can see that both the charts are giving me the same result.

256
00:22:13,080 --> 00:22:17,550
This are just two different approaches of plotting the same plot.

257
00:22:19,330 --> 00:22:24,450
Now, the last part of data visualization is autocorrelation plots.

258
00:22:25,950 --> 00:22:30,300
So if you noticed, I was talking about correlation between the leg values.

259
00:22:33,010 --> 00:22:34,530
For our forecasting model.

260
00:22:34,690 --> 00:22:39,440
It is very important to identify correlation between the lag variables.

261
00:22:40,750 --> 00:22:47,310
But SAPOL, if I want to see the correlation between this value and with the lag too well.

262
00:22:47,920 --> 00:22:55,360
So I want to see the coordination between my values and the values that the two time period ago.

263
00:22:57,070 --> 00:23:00,660
Using this method, I have to create another variable lag too.

264
00:23:00,910 --> 00:23:03,880
And then again, Lortel scatterplot on that data.

265
00:23:06,040 --> 00:23:07,810
So it's a very lengthy process.

266
00:23:07,870 --> 00:23:15,610
And that's why there is another alternative to directly get correlation values with all the possible

267
00:23:15,610 --> 00:23:16,510
leg values.

268
00:23:18,870 --> 00:23:25,170
We call that block autocorrelation plot, and this is available and fund us also.

269
00:23:25,910 --> 00:23:30,860
So we will be importing auto correlation plot from point does not plotting.

270
00:23:32,130 --> 00:23:39,850
And then we just straight autocorrelation plot and will mention the values for which even though autocorrelation

271
00:23:39,850 --> 00:23:40,060
plot.

272
00:23:40,370 --> 00:23:42,600
So if you wanted for Miles M-m.

273
00:23:44,190 --> 00:23:48,390
Let's look at the chart here.

274
00:23:48,600 --> 00:23:54,210
You can see we have cool relation values on the Y axis.

275
00:23:54,780 --> 00:23:59,400
And then we have the leg values on my x axis.

276
00:24:00,030 --> 00:24:10,700
So the first value here is the correlation value between variable miles M-m and X leg one.

277
00:24:11,940 --> 00:24:16,670
The second value is the correlation value between my DMM.

278
00:24:17,100 --> 00:24:18,830
And it's like two.

279
00:24:19,440 --> 00:24:25,860
Similarly, we have this square relation data with up to nine BD legs values.

280
00:24:28,550 --> 00:24:35,810
So now you don't have to create another lagged variables and create a scatterplot for individual lagged

281
00:24:35,810 --> 00:24:36,390
variables.

282
00:24:36,950 --> 00:24:43,520
You can just look at the single graph and this graph will tell you what is the correlation between the

283
00:24:43,520 --> 00:24:47,630
value and its lack of value at an interval.

284
00:24:49,910 --> 00:24:52,040
So now how to read this graph?

285
00:24:52,700 --> 00:24:58,700
The important thing here is from all that, we have a very high correlation for the initial three to

286
00:24:58,700 --> 00:24:59,420
four values.

287
00:24:59,630 --> 00:25:03,620
So we are getting a correlation value of more than point five.

288
00:25:05,810 --> 00:25:14,360
That mean for the first three predict, the leg values are highly correlated with the minus.

289
00:25:14,390 --> 00:25:14,990
I mean value.

290
00:25:15,680 --> 00:25:19,310
And then we have a dip at every sixth month.

291
00:25:20,750 --> 00:25:24,560
So this is this probably is the sixth or seventh value.

292
00:25:24,920 --> 00:25:27,930
This probably is the 18 or 19 value.

293
00:25:28,070 --> 00:25:30,710
Similarly, this is around 30.

294
00:25:32,360 --> 00:25:37,520
So at the multiple of every six month, we are getting a dip in our correlation values.

295
00:25:37,790 --> 00:25:39,740
This is because of the seasonality.

296
00:25:41,240 --> 00:25:45,530
So if you compare the ice cream sales over the.

297
00:25:47,360 --> 00:25:54,710
So six month correlation means when there is a summer, you are looking at the value of winter sales

298
00:25:54,860 --> 00:25:56,810
because of the six month delay.

299
00:25:57,560 --> 00:26:01,850
And when there is a winter, you are looking at the lagged value of six months.

300
00:26:01,910 --> 00:26:03,350
That is the summer values.

301
00:26:03,830 --> 00:26:09,170
That's why you will find lows and highs in case of seasonal data.

302
00:26:10,640 --> 00:26:13,340
So autocorrelation thoughts are very important.

303
00:26:13,590 --> 00:26:14,750
We'll look at.

304
00:26:14,870 --> 00:26:22,280
It's important in the later part of this course and it is very easy to plot any autocorrelation plot.

305
00:26:22,760 --> 00:26:25,640
So that's all for data visualization.

306
00:26:26,360 --> 00:26:33,500
We look at that flirting, simple line check using our data frame.

307
00:26:34,100 --> 00:26:41,360
Then we'll look at how to zoom in at any particular time interval and plot a line chart for that time

308
00:26:41,360 --> 00:26:41,840
interval.

309
00:26:42,620 --> 00:26:45,170
Then we'll look at that whole two part trend line.

310
00:26:46,970 --> 00:26:51,000
Then we look at how to move seasonality from our data.

311
00:26:51,050 --> 00:26:59,420
By aggregating data on an eleven and after that, look at how to plot scatter plots with the lagged

312
00:26:59,420 --> 00:27:06,710
values and how to plot or target a nation plot to get correlation values for all the possible land values.

313
00:27:08,420 --> 00:27:09,710
That's all for this lecture.

314
00:27:10,010 --> 00:27:10,460
Thank you.
