abdullah's picture
Add files using upload-large-folder tool
50deadf verified
raw
history blame
59.5 kB
1
00:00:17,750 --> 00:00:21,350
So let's again go back to chapter number one. Last
2
00:00:21,350 --> 00:00:25,730
time we discussed chapter one, production and data
3
00:00:25,730 --> 00:00:32,390
collection. And I think we described why learning
4
00:00:32,390 --> 00:00:36,510
statistics distinguish between some of these
5
00:00:36,510 --> 00:00:43,810
topics. And also we explained in details types of
6
00:00:43,810 --> 00:00:47,010
statistics and we mentioned that statistics mainly
7
00:00:47,010 --> 00:00:52,430
has two types either descriptive statistics which
8
00:00:52,430 --> 00:00:56,810
means collecting summarizing and obtaining data
9
00:00:56,810 --> 00:00:59,910
and other type of statistics is called inferential
10
00:00:59,910 --> 00:01:04,430
statistics or statistical inference and this type
11
00:01:04,430 --> 00:01:11,070
of statistics we can draw drawing conclusions and
12
00:01:11,070 --> 00:01:14,090
making decision concerning a population based only
13
00:01:14,090 --> 00:01:17,510
on a sample. That means we have a sample and
14
00:01:17,510 --> 00:01:20,970
sample is just a subset of the population or the
15
00:01:20,970 --> 00:01:26,230
portion of the population and we use the data from
16
00:01:26,230 --> 00:01:29,130
that sample to make some conclusion about the
17
00:01:29,130 --> 00:01:32,390
entire population. This type of statistic is
18
00:01:32,390 --> 00:01:34,710
called inferential statistics. Later, Inshallah,
19
00:01:34,750 --> 00:01:37,710
we'll talk in details about inferential statistics
20
00:01:37,710 --> 00:01:45,290
that will start in Chapter 7. Also, we gave some
21
00:01:45,290 --> 00:01:50,630
definitions for variables, data, and we
22
00:01:50,630 --> 00:01:53,510
distinguished between population and sample. And
23
00:01:53,510 --> 00:01:56,630
we know that the population consists of all items
24
00:01:56,630 --> 00:02:00,270
or individuals about which you want to draw a
25
00:02:00,270 --> 00:02:05,770
conclusion. But in some cases, it's very hard to
26
00:02:05,770 --> 00:02:07,750
talk about the population or the entire
27
00:02:07,750 --> 00:02:13,340
population, so we can select a sample. A sample is
28
00:02:13,340 --> 00:02:18,480
just a portion or subset of the entire population.
29
00:02:19,060 --> 00:02:21,860
So we know now the definition of population and
30
00:02:21,860 --> 00:02:25,360
sample. The other two types, parameter and
31
00:02:25,360 --> 00:02:28,860
statistics. Parameter is a numerical measure that
32
00:02:28,860 --> 00:02:32,300
describes characteristics of a population, while
33
00:02:32,300 --> 00:02:36,000
on the other hand, a sample, a statistic is just
34
00:02:36,430 --> 00:02:39,730
numerical measures that describe characteristic of
35
00:02:39,730 --> 00:02:44,930
a sample. So parameter is computed from the
36
00:02:44,930 --> 00:02:48,930
population while statistic is computed from the
37
00:02:48,930 --> 00:02:54,030
sample. I think we stopped at this point. Why
38
00:02:54,030 --> 00:02:56,770
collect data? I mean what are the reasons for
39
00:02:59,580 --> 00:03:01,980
One of these reasons, for example, a marketing
40
00:03:01,980 --> 00:03:04,660
research analyst needs to assess the effectiveness
41
00:03:04,660 --> 00:03:07,700
of a new television advertisement. For example,
42
00:03:07,840 --> 00:03:13,380
suppose you are a manager and you want to increase
43
00:03:13,380 --> 00:03:18,060
your salaries or your sales. Now, sales may be
44
00:03:18,060 --> 00:03:23,380
affected by advertising. So I mean, if you spend
45
00:03:23,380 --> 00:03:26,320
more on advertising, it means your sales becomes
46
00:03:26,320 --> 00:03:29,740
larger and larger. So you want to know if this
47
00:03:29,740 --> 00:03:34,160
variable, I mean if advertisement is an effective
48
00:03:34,160 --> 00:03:38,900
variable that maybe increase your sales. So that's
49
00:03:38,900 --> 00:03:43,900
one of the reasons why we use data. The other one,
50
00:03:44,120 --> 00:03:46,880
for example, pharmaceutical manufacturers needs to
51
00:03:46,880 --> 00:03:49,800
determine whether a new drug is more effective
52
00:03:49,800 --> 00:03:53,240
than those currently used. For example, for a
53
00:03:53,240 --> 00:03:59,330
headache, we use drug A. Now, a new drug is
54
00:03:59,330 --> 00:04:04,510
produced and you want to see if this new drug is
55
00:04:04,510 --> 00:04:10,090
more effective than drug A that I mean if headache
56
00:04:10,090 --> 00:04:13,410
suppose for example is removed after three days by
57
00:04:13,410 --> 00:04:20,490
using drug A now the question is does B is more
58
00:04:20,490 --> 00:04:23,410
effective it means it reduces your headache in
59
00:04:23,410 --> 00:04:26,070
fewer than three days I mean maybe in two days
60
00:04:26,510 --> 00:04:30,810
That means a drug B is more effective than a drug
61
00:04:30,810 --> 00:04:34,510
A. So we want to know the difference between these
62
00:04:34,510 --> 00:04:37,250
two drugs. I mean, we have two samples. Some
63
00:04:37,250 --> 00:04:40,810
people used drug A and the other used drug B. And
64
00:04:40,810 --> 00:04:43,190
we want to see if there is a significant
65
00:04:43,190 --> 00:04:47,690
difference between the times that is used to
66
00:04:47,690 --> 00:04:51,150
reduce the headache. So that's one of the reasons
67
00:04:51,150 --> 00:04:55,260
why we use statistics. Sometimes an operation
68
00:04:55,260 --> 00:04:58,500
manager wants to monitor manufacturing process to
69
00:04:58,500 --> 00:05:00,720
find out whether the quality of a product being
70
00:05:00,720 --> 00:05:02,840
manufactured is conforming to a company's
71
00:05:02,840 --> 00:05:05,700
standards. Do you know what the meaning of
72
00:05:05,700 --> 00:05:06,520
company's standards?
73
00:05:09,900 --> 00:05:15,320
The regulations of the firm itself. Another
74
00:05:15,320 --> 00:05:21,020
example, suppose here in the school last year, we
75
00:05:21,020 --> 00:05:25,150
teach statistics by using method A. traditional
76
00:05:25,150 --> 00:05:29,350
method. This year we developed a new method for
77
00:05:29,350 --> 00:05:32,370
teaching and our goal is to see if the new method
78
00:05:32,370 --> 00:05:36,510
is better than method A which was used in last
79
00:05:36,510 --> 00:05:38,910
year. So we want to see if there is a big
80
00:05:38,910 --> 00:05:42,410
difference between scores or the average scores
81
00:05:42,410 --> 00:05:47,310
last year and this year. The same you can do for
82
00:05:47,310 --> 00:05:52,350
your weight. Suppose there are 20 students in this
83
00:05:52,350 --> 00:05:56,960
class and their weights are high. And our goal is
84
00:05:56,960 --> 00:06:04,280
to reduce their weights. Suppose they
85
00:06:04,280 --> 00:06:09,640
have a regime or diet for three months or
86
00:06:09,640 --> 00:06:12,140
exercise, whatever it is, then after three months,
87
00:06:12,220 --> 00:06:17,060
we have new weights for these persons. And we want
88
00:06:17,060 --> 00:06:19,840
to see if the diet is effective. I mean, if the
89
00:06:19,840 --> 00:06:24,120
average weight was greater than or smaller than
90
00:06:24,120 --> 00:06:28,600
before diet. Is it clear? So there are many, many
91
00:06:28,600 --> 00:06:31,920
reasons behind using statistics and collecting
92
00:06:31,920 --> 00:06:37,500
data. Now, what are the sources of data? Since
93
00:06:37,500 --> 00:06:41,840
statistics mainly, first step, we have to collect
94
00:06:41,840 --> 00:06:44,120
data. Now, what are the sources of the data?
95
00:06:45,360 --> 00:06:48,420
Generally speaking, there are two sources. One is
96
00:06:48,420 --> 00:06:52,430
called The primary sources and the others
97
00:06:52,430 --> 00:06:55,770
secondary sources. What do you think is the
98
00:06:55,770 --> 00:06:57,830
difference between these two? I mean, what's the
99
00:06:57,830 --> 00:07:02,730
difference between primary and secondary sources?
100
00:07:03,510 --> 00:07:07,250
The primary source is the collector of the data.
101
00:07:07,670 --> 00:07:11,550
He is the analyzer. He analyzes it. And then the
102
00:07:11,550 --> 00:07:14,230
secondary, who collects the data, isn't there.
103
00:07:16,030 --> 00:07:18,910
That's correct. So the primary sources means the
104
00:07:18,910 --> 00:07:22,490
researcher by himself. He should collect the data,
105
00:07:23,890 --> 00:07:27,750
then he can use the data to do his analysis.
106
00:07:28,310 --> 00:07:31,550
That's for the primary. Now, the primary could be
107
00:07:31,550 --> 00:07:35,230
data from political survey. You can distribute
108
00:07:35,230 --> 00:07:38,750
questionnaire, for example, data collected from an
109
00:07:38,750 --> 00:07:42,530
experiment. I mean maybe control or experimental
110
00:07:42,530 --> 00:07:45,730
groups. We have two groups, maybe healthy people
111
00:07:45,730 --> 00:07:48,490
and patient people. So that's experimental group.
112
00:07:49,010 --> 00:07:53,390
Or observed data. That's the primary sources.
113
00:07:53,870 --> 00:07:56,450
Secondary sources, the person performing data
114
00:07:56,450 --> 00:08:00,310
analysis is not the data collector. So he obtained
115
00:08:00,310 --> 00:08:03,880
the data from other sources. For example, it could
116
00:08:03,880 --> 00:08:07,140
be analyzing census data or for example, examining
117
00:08:07,140 --> 00:08:10,160
data from print journals or data published on the
118
00:08:10,160 --> 00:08:14,780
internet. So maybe he goes to the Ministry of
119
00:08:14,780 --> 00:08:18,820
Education and he can get some data. So the data is
120
00:08:18,820 --> 00:08:22,520
already there and he just used the data to do some
121
00:08:22,520 --> 00:08:25,540
analysis. So that's the difference between a
122
00:08:25,540 --> 00:08:29,420
primary and secondary sources. So primary, the
123
00:08:29,420 --> 00:08:33,650
researcher himself, should collect the data by
124
00:08:33,650 --> 00:08:35,590
using one of the tools, either survey,
125
00:08:36,110 --> 00:08:39,050
questionnaire, experiment, and so on. But
126
00:08:39,050 --> 00:08:41,450
secondary, you can use the data that is published
127
00:08:41,450 --> 00:08:44,510
in the internet, for example, in the books, in
128
00:08:44,510 --> 00:08:48,250
governments and NGOs and so on. So these are the
129
00:08:48,250 --> 00:08:53,590
two sources of data. Sources of data fall into
130
00:08:53,590 --> 00:08:57,610
four categories. Number one, data distributed by
131
00:08:57,610 --> 00:09:01,190
an organization or an individual. So that's
132
00:09:01,190 --> 00:09:06,170
secondary source. A design experiment is primary
133
00:09:06,170 --> 00:09:10,350
because you have to design the experiment, a
134
00:09:10,350 --> 00:09:14,610
survey. It's also primary. An observational study
135
00:09:14,610 --> 00:09:17,590
is also a primary source. So you have to
136
00:09:17,590 --> 00:09:21,410
distinguish between a primary and secondary
137
00:09:21,410 --> 00:09:28,810
sources. Any question? Comments? Next.
138
00:09:31,460 --> 00:09:34,540
We'll talk a little bit about types of variables.
139
00:09:35,360 --> 00:09:37,580
In general, there are two types of variables. One
140
00:09:37,580 --> 00:09:40,760
is called categorical variables or qualitative
141
00:09:40,760 --> 00:09:43,160
variables, and the other one is called numerical
142
00:09:43,160 --> 00:09:46,520
or quantitative variables. Now, for example, if I
143
00:09:46,520 --> 00:09:50,560
ask you, what's
144
00:09:50,560 --> 00:09:55,100
your favorite color? You may say white, black,
145
00:09:55,220 --> 00:09:59,390
red, and so on. What's your marital status? Maybe
146
00:09:59,390 --> 00:10:02,670
married or unmarried, and so on. Gender, male,
147
00:10:02,850 --> 00:10:07,050
either male or female, and so on. So this type of
148
00:10:07,050 --> 00:10:10,090
variable is called qualitative variables. So
149
00:10:10,090 --> 00:10:13,130
qualitative variables have values that can only be
150
00:10:13,130 --> 00:10:16,370
placed into categories, such as, for example, yes
151
00:10:16,370 --> 00:10:21,350
or no. For example, do you like orange?
152
00:10:22,270 --> 00:10:26,200
The answer is either yes or no. Do you like
153
00:10:26,200 --> 00:10:30,040
candidate A, for example, whatever his party is?
154
00:10:30,260 --> 00:10:34,620
Do you like it? Either yes or no, and so on. As I
155
00:10:34,620 --> 00:10:37,480
mentioned before, gender, marital status, race,
156
00:10:37,640 --> 00:10:41,820
religions, these are examples of qualitative or
157
00:10:41,820 --> 00:10:46,240
categorical variables. The other type of variable
158
00:10:46,240 --> 00:10:49,480
which is commonly used is called numerical or
159
00:10:49,480 --> 00:10:53,230
quantitative data. Quantitative variables have
160
00:10:53,230 --> 00:10:56,350
values that represent quantities. For example, if
161
00:10:56,350 --> 00:11:00,470
I ask you, what's your age? My age is 20 years old
162
00:11:00,470 --> 00:11:04,770
or 18 years old. What's your weight? Income.
163
00:11:07,510 --> 00:11:12,550
Height? Temperature? Income. So it's a number.
164
00:11:13,610 --> 00:11:18,030
Weight, maybe my weight is 70 kilograms. So
165
00:11:18,030 --> 00:11:22,450
weight, age, height, salary, income, number of
166
00:11:22,450 --> 00:11:26,510
students, number of phone calls you received on
167
00:11:26,510 --> 00:11:28,770
your cell phone during one hour, number of
168
00:11:28,770 --> 00:11:33,210
accidents happened in street and so on. So that's
169
00:11:33,210 --> 00:11:36,470
the difference between numerical variables and
170
00:11:36,470 --> 00:11:37,710
qualitative variables.
171
00:11:40,270 --> 00:11:42,650
Anyone of you just give me one example of
172
00:11:42,650 --> 00:11:47,780
qualitative and quantitative variables. Another
173
00:11:47,780 --> 00:11:51,700
examples. Just give me one example for qualitative
174
00:11:51,700 --> 00:11:52,160
data.
175
00:11:56,900 --> 00:11:58,780
Qualitative or quantitative.
176
00:12:01,380 --> 00:12:04,880
Political party, either party A or party B. So
177
00:12:04,880 --> 00:12:08,320
suppose there are two parties, so I like party A,
178
00:12:08,720 --> 00:12:12,320
she likes party B and so on. So party in this case
179
00:12:12,320 --> 00:12:15,060
is qualitative variable, another one.
180
00:12:25,400 --> 00:12:28,820
So types of courses, maybe business, economics,
181
00:12:29,480 --> 00:12:33,260
administration, and so on. So types of courses.
182
00:12:34,260 --> 00:12:36,600
Another example for quantitative variable or
183
00:12:36,600 --> 00:12:37,600
numerical variables.
184
00:12:45,020 --> 00:12:51,440
So production is a numerical variable. Another
185
00:12:51,440 --> 00:12:52,840
example for quantitative.
186
00:12:56,350 --> 00:12:58,970
Is that produced by this company? Number of cell
187
00:12:58,970 --> 00:13:03,950
phones, maybe 20, 17, and so on. Any question?
188
00:13:06,190 --> 00:13:12,410
Next. So generally speaking, types of data, data
189
00:13:12,410 --> 00:13:17,960
has two types, categorical and numerical data. As
190
00:13:17,960 --> 00:13:21,240
we mentioned, marital status, political party, eye
191
00:13:21,240 --> 00:13:25,120
color, and so on. These are examples of
192
00:13:25,120 --> 00:13:28,180
categorical variables. On the other hand, a
193
00:13:28,180 --> 00:13:30,720
numerical variable can be split or divided into
194
00:13:30,720 --> 00:13:34,020
two parts. One is called discrete and the other is
195
00:13:34,020 --> 00:13:35,680
continuous, and we have to distinguish between
196
00:13:35,680 --> 00:13:40,240
these two. For example, Number of students in this
197
00:13:40,240 --> 00:13:43,400
class, you can say there are 60 or 50 students in
198
00:13:43,400 --> 00:13:47,260
this class. You cannot say there are 50.5
199
00:13:47,260 --> 00:13:52,160
students. So number of students is discrete
200
00:13:52,160 --> 00:13:57,040
because it takes only integers. While for
201
00:13:57,040 --> 00:13:59,800
continuous type of numerical variables, you can
202
00:13:59,800 --> 00:14:06,060
say that my weight is 80.5 kilograms. so it makes
203
00:14:06,060 --> 00:14:09,260
sense that your weight is not exactly 80 kilograms
204
00:14:09,260 --> 00:14:15,580
it might be 80.6 or 80.5 and so on so discrete
205
00:14:15,580 --> 00:14:20,420
takes only integers while continuous takes any
206
00:14:20,420 --> 00:14:24,140
value I mean any real number so that's the
207
00:14:24,140 --> 00:14:28,980
difference between discrete and continuous number
208
00:14:28,980 --> 00:14:33,930
of phone or number of calls you have received this
209
00:14:33,930 --> 00:14:37,670
morning, maybe one, zero, nine, and so on,
210
00:14:38,390 --> 00:14:41,950
discrete. Number of patients in the hospital,
211
00:14:42,350 --> 00:14:46,590
discrete, and so on. But when we are talking about
212
00:14:46,590 --> 00:14:50,850
income, maybe my income is 1,000.5 shekel. It
213
00:14:50,850 --> 00:14:55,790
could be. It's continuous because my income can be
214
00:14:55,790 --> 00:14:59,210
any number between, for example, 1,000 and 10,000.
215
00:15:00,470 --> 00:15:03,550
It takes any value in this interval from 1,000 to
216
00:15:03,550 --> 00:15:08,410
10,000. So it types of continuous rather than our
217
00:15:08,410 --> 00:15:12,450
continuous variable. So that's the two types of
218
00:15:12,450 --> 00:15:15,590
data, categorical and numerical. And numerical
219
00:15:15,590 --> 00:15:18,510
also has two types, either discrete or continuous.
220
00:15:19,750 --> 00:15:24,430
Later in Chapter 6, we'll talk more details about
221
00:15:24,430 --> 00:15:28,940
one of the most distribution statistics, for
222
00:15:28,940 --> 00:15:31,560
continuous, one which is called normal
223
00:15:31,560 --> 00:15:36,240
distribution. That will be later, inshallah. As we
224
00:15:36,240 --> 00:15:39,000
mentioned last time, at the end of each chapter,
225
00:15:39,280 --> 00:15:41,640
there is a section or sections, sometimes there
226
00:15:41,640 --> 00:15:45,000
are two sections, talks about computer programs.
227
00:15:45,120 --> 00:15:49,200
How can we use computer programs in order to
228
00:15:49,200 --> 00:15:52,300
analyze the data? And as we mentioned last time,
229
00:15:52,420 --> 00:15:54,600
you should take a course on that. It's called
230
00:15:54,600 --> 00:15:57,960
Computer and Data Analysis or SPSS course. So we
231
00:15:57,960 --> 00:16:03,860
are going to skip the computer programs used for
232
00:16:03,860 --> 00:16:08,720
any chapters in this book. And that's the end of
233
00:16:08,720 --> 00:16:13,060
chapter number three. Any questions?
234
00:16:18,380 --> 00:16:22,550
Let's move. quickly on chapter three.
235
00:16:31,290 --> 00:16:35,530
Chapter three maybe is the easiest chapter in this
236
00:16:35,530 --> 00:16:39,950
book. It's straightforward. We have some formulas
237
00:16:39,950 --> 00:16:45,280
to compute some statistical measures. And we
238
00:16:45,280 --> 00:16:47,400
should know how can we calculate these measures
239
00:16:47,400 --> 00:16:52,620
and what are the meaning of your results. So
240
00:16:52,620 --> 00:16:56,140
chapter three talks about numerical descriptive
241
00:16:56,140 --> 00:17:03,220
measures. In this chapter, you will learn, number
242
00:17:03,220 --> 00:17:06,480
one, describe the probabilities of central
243
00:17:06,480 --> 00:17:11,880
tendency, variation, and shape in numerical data.
244
00:17:12,730 --> 00:17:16,250
In this lecture, we'll talk in more details about
245
00:17:16,250 --> 00:17:21,370
some of the center tendency measures. Later, we'll
246
00:17:21,370 --> 00:17:26,490
talk about the variation, or spread, or
247
00:17:26,490 --> 00:17:29,690
dispersion, and the shape in numerical data. So
248
00:17:29,690 --> 00:17:31,630
that's part number one. We have to know something
249
00:17:31,630 --> 00:17:36,390
about center tendency, variation, and the shape of
250
00:17:36,390 --> 00:17:42,020
the data we have. to calculate descriptive summary
251
00:17:42,020 --> 00:17:45,360
measures for a population. So we have to calculate
252
00:17:45,360 --> 00:17:48,460
these measures for the sample. And if we have the
253
00:17:48,460 --> 00:17:51,420
entire population, we can compute these measures
254
00:17:51,420 --> 00:17:58,440
also for that population. Then I will introduce in
255
00:17:58,440 --> 00:18:01,920
more details about something called Paxiplot. How
256
00:18:01,920 --> 00:18:06,660
can we construct and interpret a Paxiplot? That's,
257
00:18:06,960 --> 00:18:11,040
inshallah, next time on Tuesday. Finally, we'll
258
00:18:11,040 --> 00:18:13,020
see how can we calculate the covariance and
259
00:18:13,020 --> 00:18:15,700
coefficient of variation and coefficient, I'm
260
00:18:15,700 --> 00:18:18,480
sorry, coefficient of correlation. This topic
261
00:18:18,480 --> 00:18:24,280
we'll introduce in more details in chapter 11
262
00:18:24,280 --> 00:18:31,240
later on. So just I will give some brief
263
00:18:31,240 --> 00:18:35,630
notation about coefficient of correlation, how can
264
00:18:35,630 --> 00:18:39,030
we compute the correlation coefficient? What's the
265
00:18:39,030 --> 00:18:41,870
meaning of your result? And later in chapter 11,
266
00:18:42,030 --> 00:18:44,510
we'll talk in more details about correlation and
267
00:18:44,510 --> 00:18:48,930
regression. So these are the objectives of this
268
00:18:48,930 --> 00:18:52,870
chapter. There are some basic definitions before
269
00:18:52,870 --> 00:18:57,410
we start. One is called central tendency. What do
270
00:18:57,410 --> 00:19:00,750
you mean by central tendency? Central tendency is
271
00:19:00,750 --> 00:19:04,990
the extent to which all data value group around a
272
00:19:04,990 --> 00:19:08,890
typical or numerical or central value. So we are
273
00:19:08,890 --> 00:19:12,510
looking for a point that in the center, I mean,
274
00:19:12,810 --> 00:19:18,870
the data points are gathered or collected around a
275
00:19:18,870 --> 00:19:21,670
middle point, and that middle point is called the
276
00:19:21,670 --> 00:19:24,450
central tendency. And the question is, how can we
277
00:19:24,450 --> 00:19:27,780
measure that value? We'll talk in details about
278
00:19:27,780 --> 00:19:32,620
mean, median, and mode in a few minutes. So the
279
00:19:32,620 --> 00:19:35,660
central tendency, in this case, the data values
280
00:19:35,660 --> 00:19:40,080
grouped around a typical or central value. Is it
281
00:19:40,080 --> 00:19:44,380
clear? So we have data set, large data set. Then
282
00:19:44,380 --> 00:19:47,860
these points are gathered or grouped around a
283
00:19:47,860 --> 00:19:51,440
middle point, and this point is called central
284
00:19:51,440 --> 00:19:56,120
tendency, and it can be measured by using mean,
285
00:19:56,420 --> 00:19:59,960
which is the most common one, median and the moon.
286
00:20:01,020 --> 00:20:04,480
Next is the variation, which is the amount of
287
00:20:04,480 --> 00:20:09,420
dispersion. Variation is the amount of dispersion
288
00:20:09,420 --> 00:20:13,900
or scattering of values. And we'll use, for
289
00:20:13,900 --> 00:20:18,400
example, range, variance or standard deviation in
290
00:20:18,400 --> 00:20:22,960
order to compute the variation. Finally, We have
291
00:20:22,960 --> 00:20:26,300
data, and my question is, what's the shape of the
292
00:20:26,300 --> 00:20:29,920
data? So the shape is the pattern of distribution
293
00:20:29,920 --> 00:20:35,220
of values from the lowest value to the highest. So
294
00:20:35,220 --> 00:20:39,400
that's the three definitions we need to know
295
00:20:39,400 --> 00:20:44,580
before we start. So we'll start with the easiest
296
00:20:44,580 --> 00:20:48,680
one, measures of central tendency. As I mentioned,
297
00:20:49,160 --> 00:20:55,110
there are three measures. median and moon. And our
298
00:20:55,110 --> 00:20:58,270
goal or we have two goals actually. We have to
299
00:20:58,270 --> 00:21:02,290
know how to compute these measures. Number two,
300
00:21:03,270 --> 00:21:06,390
which one is better? The mean or the median or the
301
00:21:06,390 --> 00:21:06,550
moon?
302
00:21:11,310 --> 00:21:14,770
So the mean sometimes called the arithmetic mean.
303
00:21:15,680 --> 00:21:20,020
Or in general, just say the mean. So often we use
304
00:21:20,020 --> 00:21:26,860
the mean. And the mean is just sum
305
00:21:26,860 --> 00:21:33,220
of the values divided by the sample size. So it's
306
00:21:33,220 --> 00:21:36,800
straightforward. We have, for example, three data
307
00:21:36,800 --> 00:21:42,180
points. And your goal is to find the average or
308
00:21:42,180 --> 00:21:45,890
the mean of these points. They mean it's just some
309
00:21:45,890 --> 00:21:50,230
of these values divided by the sample size. So for
310
00:21:50,230 --> 00:21:54,570
example, if we have a data X1, X2, X3 up to Xn. So
311
00:21:54,570 --> 00:21:59,650
the average is denoted by X bar. This one is
312
00:21:59,650 --> 00:22:04,530
pronounced as X bar and X bar is just sum of Xi.
313
00:22:05,010 --> 00:22:08,250
It is summation, you know this symbol, summation
314
00:22:08,250 --> 00:22:11,350
of sigma, summation of Xi and I goes from one to
315
00:22:11,350 --> 00:22:14,490
N. divided by N which is the total number of
316
00:22:14,490 --> 00:22:19,710
observations or the sample size. So it means X1
317
00:22:19,710 --> 00:22:23,290
plus X2 all the way up to XN divided by N gives
318
00:22:23,290 --> 00:22:28,530
the mean or the arithmetic mean. So X bar is the
319
00:22:28,530 --> 00:22:32,690
average which is the sum of values divided by the
320
00:22:32,690 --> 00:22:36,270
number of observations. So that's the first
321
00:22:36,270 --> 00:22:38,830
definition. For example,
322
00:22:42,180 --> 00:22:46,920
So again, the mean is the most common measure of
323
00:22:46,920 --> 00:22:51,780
center tendency. Number two, the definition of the
324
00:22:51,780 --> 00:22:55,440
mean. Sum of values divided by the number of
325
00:22:55,440 --> 00:23:02,960
values. That means the mean takes all the values,
326
00:23:04,140 --> 00:23:09,740
then divided by N. it makes sense that the mean is
327
00:23:09,740 --> 00:23:13,380
affected by extreme values or outliers. I mean, if
328
00:23:13,380 --> 00:23:17,840
the data has outliers or extreme values, I mean by
329
00:23:17,840 --> 00:23:21,400
extreme values, large or very, very large values
330
00:23:21,400 --> 00:23:24,980
and small, small values. Large values or small
331
00:23:24,980 --> 00:23:31,100
values are extreme values. Since the mean takes
332
00:23:31,100 --> 00:23:33,420
all these values and sums all together, doesn't
333
00:23:33,420 --> 00:23:38,550
divide by n, that means The mean is affected by
334
00:23:38,550 --> 00:23:41,350
outliers or by extreme values. For example,
335
00:23:42,030 --> 00:23:45,110
imagine we have simple data as 1, 2, 3, 4, and 5.
336
00:23:46,110 --> 00:23:49,830
Simple example. Now, what's the mean? The mean is
337
00:23:49,830 --> 00:23:53,570
just add these values, then divide by the total
338
00:23:53,570 --> 00:23:56,910
number of observations. In this case, the sum of
339
00:23:56,910 --> 00:24:01,710
these is 15. N is five because there are five
340
00:24:01,710 --> 00:24:05,920
observations. So X bar is 15 divided by 5, which
341
00:24:05,920 --> 00:24:10,240
is 3. So straightforward. Now imagine instead of
342
00:24:10,240 --> 00:24:16,480
5, this number 5, we have a 10. Now 10, there is a
343
00:24:16,480 --> 00:24:21,400
gap between 4, which is the second largest, and
344
00:24:21,400 --> 00:24:25,600
the maximum, which is 10. Now if we add these
345
00:24:25,600 --> 00:24:30,540
values, 1, 2, 3, 4, and 10, then divide by 5, the
346
00:24:30,540 --> 00:24:36,680
mean will be 4. If you see here, we just added one
347
00:24:36,680 --> 00:24:41,060
value, or I mean, we replaced five by 10, and the
348
00:24:41,060 --> 00:24:44,700
mean changed dramatically from three to four.
349
00:24:45,520 --> 00:24:48,860
There is big change between three and four, around
350
00:24:48,860 --> 00:24:55,560
25% more. So that means outliers or extreme values
351
00:24:55,560 --> 00:25:01,200
affected the mean. So take this information in
352
00:25:01,200 --> 00:25:03,560
your mind because later we'll talk a little bit
353
00:25:03,560 --> 00:25:07,360
about another one. So the mean is affected by
354
00:25:07,360 --> 00:25:13,100
extreme values. Imagine another example. Suppose
355
00:25:13,100 --> 00:25:20,060
we have data from 1 to 9. 1, 2, 3, 4, 6, 7, 8, 9.
356
00:25:21,040 --> 00:25:26,690
Now the mean of these values, some divide by n. If
357
00:25:26,690 --> 00:25:31,970
you sum 1 through 9, summation is 45. Divide by 9,
358
00:25:32,510 --> 00:25:36,230
which is 5. So the sum of these values divided by
359
00:25:36,230 --> 00:25:41,590
N gives the average, so the average is 5. Now
360
00:25:41,590 --> 00:25:46,670
suppose we add 100 to the end of this data. So the
361
00:25:46,670 --> 00:25:53,670
sum will be 145 divided by 10, that's 14.5. Now
362
00:25:53,670 --> 00:25:58,850
the mean was 5. Then after we added 100, it
363
00:25:58,850 --> 00:26:05,470
becomes 14.5. Imagine the mean was 5, it changed
364
00:26:05,470 --> 00:26:11,650
to 14.5. It means around three times. So that
365
00:26:11,650 --> 00:26:17,510
means outliers affect the mean much more than the
366
00:26:17,510 --> 00:26:19,890
other one. We'll talk a little later about it,
367
00:26:19,990 --> 00:26:23,950
which is the median. So keep in mind outliers
368
00:26:25,290 --> 00:26:34,790
affected the mean in this case. Any question? Is
369
00:26:34,790 --> 00:26:41,590
it clear? Yes. So, one more time. The mean is
370
00:26:41,590 --> 00:26:46,990
affected by extreme values. So that's for the
371
00:26:46,990 --> 00:26:50,910
mean. The other measure of center tendency is
372
00:26:50,910 --> 00:26:57,600
called the median. Now, what's the median? What's
373
00:26:57,600 --> 00:27:00,760
the definition of the median from your previous
374
00:27:00,760 --> 00:27:05,880
studies? What's the median? I mean, what's the
375
00:27:05,880 --> 00:27:09,360
definition of the median? Now the middle value,
376
00:27:09,760 --> 00:27:12,980
that's correct, but after we arrange the data from
377
00:27:12,980 --> 00:27:17,040
smallest to largest or largest to smallest, so we
378
00:27:17,040 --> 00:27:20,160
should arrange the data, then we can figure out
379
00:27:20,160 --> 00:27:24,280
the median. So the median is the middle point, but
380
00:27:24,280 --> 00:27:27,060
after we arrange the data from smallest to largest
381
00:27:27,060 --> 00:27:30,030
or vice versa. So that's the definition of the
382
00:27:30,030 --> 00:27:33,930
median. So in an ordered array, so we have to have
383
00:27:33,930 --> 00:27:39,230
order array, the median is the middle number. The
384
00:27:39,230 --> 00:27:42,810
middle number means 50 percent of the data below
385
00:27:42,810 --> 00:27:50,370
and 50 percent above the median because it's
386
00:27:50,370 --> 00:27:52,190
called the median, the value in the middle after
387
00:27:52,190 --> 00:27:55,990
you arrange the data from smallest to largest.
388
00:28:00,130 --> 00:28:02,770
Suppose I again go back to the previous example
389
00:28:02,770 --> 00:28:09,690
when we have data 1, 2, 3, 4, and 5. Now for this
390
00:28:09,690 --> 00:28:14,210
specific example as we did before, now the data is
391
00:28:14,210 --> 00:28:18,670
already ordered. The value in the middle is 3
392
00:28:18,670 --> 00:28:22,330
because there are two pillows.
393
00:28:24,860 --> 00:28:27,300
And also there are the same number of observations
394
00:28:27,300 --> 00:28:33,140
above it. So 3 is the median. Now again imagine we
395
00:28:33,140 --> 00:28:37,320
replace 5, which is the maximum value, by another
396
00:28:37,320 --> 00:28:42,140
one which is extreme one, for example 10. In this
397
00:28:42,140 --> 00:28:47,600
case, the median is still 3. Because the median is
398
00:28:47,600 --> 00:28:49,380
just the value of the middle after you arrange the
399
00:28:49,380 --> 00:28:53,900
data. So it doesn't matter what is the highest or
400
00:28:53,900 --> 00:28:58,860
the maximum value is, the median in this case is
401
00:28:58,860 --> 00:29:03,700
three. It doesn't change. That means the median is
402
00:29:03,700 --> 00:29:08,020
not affected by extreme values. Or to be more
403
00:29:08,020 --> 00:29:12,910
precise, we can say that The median is affected by
404
00:29:12,910 --> 00:29:18,990
outlier, but not the same as the mean. So affect
405
00:29:18,990 --> 00:29:23,610
the mean much more than the median. I mean, you
406
00:29:23,610 --> 00:29:26,550
cannot say for this example, yes, the median is
407
00:29:26,550 --> 00:29:29,310
not affected because the median was three, it
408
00:29:29,310 --> 00:29:33,590
becomes three. But in another examples, there is
409
00:29:33,590 --> 00:29:36,750
small difference between all.
410
00:29:40,770 --> 00:29:44,850
Extreme values affected the mean much more than
411
00:29:44,850 --> 00:29:51,450
the median. If the dataset has extreme values, we
412
00:29:51,450 --> 00:29:54,510
have to use, what do you think, the mean or the
413
00:29:54,510 --> 00:29:58,090
median? The median. So in case or in the presence
414
00:29:58,090 --> 00:30:01,910
of extreme values or outliers, we have to use the
415
00:30:01,910 --> 00:30:07,010
median, not the mean. But in general, we use If
416
00:30:07,010 --> 00:30:10,770
the data is free of outliers, I mean if the data
417
00:30:10,770 --> 00:30:16,410
has not extreme values, then you can use the mean.
418
00:30:16,510 --> 00:30:19,230
The mean is much better than the median in this
419
00:30:19,230 --> 00:30:22,490
case. But if the data has extreme values or
420
00:30:22,490 --> 00:30:27,190
outliers, we should use the median instead of the
421
00:30:27,190 --> 00:30:31,310
mean. Any question? So these are the most common
422
00:30:31,310 --> 00:30:36,710
center tendency measures in statistics, the mean
423
00:30:36,710 --> 00:30:42,390
and the median. And keep in mind, your data should
424
00:30:42,390 --> 00:30:46,170
be numeric. I mean, you cannot use the mean or the
425
00:30:46,170 --> 00:30:50,730
median for qualitative or categorical data, for
426
00:30:50,730 --> 00:30:54,310
example, gender, males or females. You cannot say
427
00:30:54,310 --> 00:30:59,490
the mean of gender or sex is 1.5. It doesn't make
428
00:30:59,490 --> 00:31:05,490
sense. It should be numerical data to use the mean
429
00:31:05,490 --> 00:31:07,590
or the median. So the mean and the median is used
430
00:31:07,590 --> 00:31:11,210
only for numerical data. And we have to
431
00:31:11,210 --> 00:31:14,170
distinguish between mean and median. Mean is used
432
00:31:14,170 --> 00:31:16,870
for data that has not outliers or extreme values,
433
00:31:17,370 --> 00:31:21,450
while the median is used for data that has
434
00:31:21,450 --> 00:31:25,230
outliers or extreme values. Sometimes better to
435
00:31:25,230 --> 00:31:27,990
report both. I mean, sometimes better to report
436
00:31:27,990 --> 00:31:33,450
mean and the median. So you just say the sales for
437
00:31:33,450 --> 00:31:40,560
this company is, for example, $500,000. And the
438
00:31:40,560 --> 00:31:43,900
median, for example, is 550,000. You can see that.
439
00:31:45,680 --> 00:31:46,400
Is it clear?
440
00:31:51,440 --> 00:31:55,560
If you have a small data, it's straightforward and
441
00:31:55,560 --> 00:31:59,180
it's very easy to locate the median. But in case
442
00:31:59,180 --> 00:32:02,120
of large dataset, how can we locate the median?
443
00:32:02,340 --> 00:32:06,640
It's not easy. Just look at the data and you can
444
00:32:06,640 --> 00:32:11,200
say this is the median. It's not easy task. So we
445
00:32:11,200 --> 00:32:15,820
need a rule that locate the median. The location
446
00:32:15,820 --> 00:32:18,020
of the median when the values are in numerical
447
00:32:18,020 --> 00:32:23,580
order from smallest to largest is N plus one
448
00:32:23,580 --> 00:32:26,140
divided by two. That's the position of the median.
449
00:32:26,640 --> 00:32:28,860
If we go back a little bit to the previous
450
00:32:28,860 --> 00:32:34,980
example, here N was five. So the location was
451
00:32:34,980 --> 00:32:40,000
number three, because n plus one divided by two,
452
00:32:40,120 --> 00:32:43,120
five plus one divided by two is three. So location
453
00:32:43,120 --> 00:32:47,340
number three is the median. Location number one is
454
00:32:47,340 --> 00:32:50,840
one, in this case, then two, then three. So
455
00:32:50,840 --> 00:32:53,740
location number three is three. But maybe this
456
00:32:53,740 --> 00:32:57,280
number is not three, and other value maybe 3.1 or
457
00:32:57,280 --> 00:33:02,440
3.2. But the location is number three. Is it
458
00:33:02,440 --> 00:33:08,470
clear? So that's the location. If it is odd, you
459
00:33:08,470 --> 00:33:13,270
mean by odd number, five, seven and so on. So if
460
00:33:13,270 --> 00:33:17,090
the number of values is odd, the median is the
461
00:33:17,090 --> 00:33:21,210
middle number. Now let's imagine if we have even
462
00:33:21,210 --> 00:33:24,570
number of observations. For example, we have one,
463
00:33:24,610 --> 00:33:28,270
two, three, four, five and six. So imagine numbers
464
00:33:28,270 --> 00:33:32,390
from one up to six. What's the median? Now three
465
00:33:32,390 --> 00:33:35,610
is not the median because there are two
466
00:33:35,610 --> 00:33:43,390
observations below three. And three above it. And
467
00:33:43,390 --> 00:33:46,210
four is not the median because three observations
468
00:33:46,210 --> 00:33:53,290
below, two above. So three and four is the middle
469
00:33:53,290 --> 00:33:56,870
value. So just take the average of two middle
470
00:33:56,870 --> 00:34:01,570
points, And that will be the median. So if n is
471
00:34:01,570 --> 00:34:07,990
even, you have to locate two middle points. For
472
00:34:07,990 --> 00:34:10,310
example, n over 2, in this case, we have six
473
00:34:10,310 --> 00:34:13,910
observations. So divide by 2, not n plus 1 divided
474
00:34:13,910 --> 00:34:17,970
by 2, just n over 2. So n over 2 is 3. So place
475
00:34:17,970 --> 00:34:22,930
number 3, and the next one, place number 4, these
476
00:34:22,930 --> 00:34:25,930
are the two middle points. Take the average of
477
00:34:25,930 --> 00:34:32,300
these values, then that's your median. So if N is
478
00:34:32,300 --> 00:34:37,080
even, you have to be careful. You have to find two
479
00:34:37,080 --> 00:34:40,860
middle points and just take the average of these
480
00:34:40,860 --> 00:34:45,100
two. So if N is even, the median is the average of
481
00:34:45,100 --> 00:34:49,200
the two middle numbers. Keep in mind, when we are
482
00:34:49,200 --> 00:34:54,600
saying N plus 2, N plus 2 is just the position of
483
00:34:54,600 --> 00:34:58,670
the median, not the value, location. Not the
484
00:34:58,670 --> 00:35:07,770
value. Is it clear? Any question? So location is
485
00:35:07,770 --> 00:35:10,150
not the value. Location is just the place or the
486
00:35:10,150 --> 00:35:13,450
position of the medium. If N is odd, the position
487
00:35:13,450 --> 00:35:17,710
is N plus one divided by two. If N is even, the
488
00:35:17,710 --> 00:35:20,870
positions of the two middle points are N over two
489
00:35:20,870 --> 00:35:23,090
and the next term or the next point.
490
00:35:28,390 --> 00:35:32,510
Last measure of center tendency is called the
491
00:35:32,510 --> 00:35:32,750
mood.
492
00:35:35,890 --> 00:35:39,010
The definition of the mood, the mood is the most
493
00:35:39,010 --> 00:35:44,250
frequent value. So sometimes the mood exists,
494
00:35:45,230 --> 00:35:48,570
sometimes the mood does not exist. Or sometimes
495
00:35:48,570 --> 00:35:53,730
there is only one mood, in other cases maybe there
496
00:35:53,730 --> 00:35:58,730
are several moods. So a value that occurs most
497
00:35:58,730 --> 00:36:03,010
often is called the mood. The mood is not affected
498
00:36:03,010 --> 00:36:07,610
by extreme values. It can be used for either
499
00:36:07,610 --> 00:36:11,190
numerical or categorical data. And that's the
500
00:36:11,190 --> 00:36:13,910
difference between mean and median and the mood.
501
00:36:14,590 --> 00:36:16,930
Mean and median is used just for numerical data.
502
00:36:17,430 --> 00:36:21,270
Here, the mood can be used for both, categorical
503
00:36:21,270 --> 00:36:25,610
and numerical data. Sometimes, as I mentioned,
504
00:36:25,930 --> 00:36:29,570
there may be no mood or the mood does not exist.
505
00:36:30,130 --> 00:36:34,190
In other cases, there may be several events. So
506
00:36:34,190 --> 00:36:36,870
the mood is the value that has the most frequent.
507
00:36:37,490 --> 00:36:43,650
For example, if you look at this data, one is
508
00:36:43,650 --> 00:36:48,370
repeated once, three is the same one time, five is
509
00:36:48,370 --> 00:36:52,290
repeated twice. seven is one nine is repeated
510
00:36:52,290 --> 00:36:57,330
three times and so on so in this case nine is the
511
00:36:57,330 --> 00:37:00,290
mood because the mood again is the most frequent
512
00:37:00,290 --> 00:37:05,030
value on
513
00:37:05,030 --> 00:37:08,550
the right side there are some values zero one two
514
00:37:08,550 --> 00:37:12,830
three up to six now each one is repeated once so
515
00:37:12,830 --> 00:37:15,350
in this case the mood does not exist I mean there
516
00:37:15,350 --> 00:37:22,310
is no mood So generally speaking, the mood is the
517
00:37:22,310 --> 00:37:26,310
value that you care most often. It can be used for
518
00:37:26,310 --> 00:37:29,790
numerical or categorical data, not affected by
519
00:37:29,790 --> 00:37:32,970
extreme values or outliers. Sometimes there is
520
00:37:32,970 --> 00:37:36,150
only one mood as this example. Sometimes the mood
521
00:37:36,150 --> 00:37:40,390
does not exist. Or sometimes there are several
522
00:37:40,390 --> 00:37:45,190
moods. And so that's the definitions for mean,
523
00:37:46,430 --> 00:37:52,540
median, and the mood. I will give just a numerical
524
00:37:52,540 --> 00:37:56,380
example to know how can we compute these measures.
525
00:37:57,420 --> 00:38:01,540
This data, simple data, just for illustration, we
526
00:38:01,540 --> 00:38:07,580
have house prices. We have five data points, $2
527
00:38:07,580 --> 00:38:10,940
million. This is the price of house A, for
528
00:38:10,940 --> 00:38:15,880
example. House B price is 500,000. The other one
529
00:38:15,880 --> 00:38:19,120
is 300,000. And two houses have the same price as
530
00:38:19,120 --> 00:38:25,850
100,000. Now, just to compute the mean, add these
531
00:38:25,850 --> 00:38:29,350
values or sum these values, which is three
532
00:38:29,350 --> 00:38:34,030
million, divide by number of houses here, there
533
00:38:34,030 --> 00:38:38,550
are five houses, so just three thousand divided by
534
00:38:38,550 --> 00:38:44,170
five, six hundred thousand. The median, the value
535
00:38:44,170 --> 00:38:46,150
in the median, after you arrange the data from
536
00:38:46,150 --> 00:38:51,470
smallest to largest, Or largest smallest. This
537
00:38:51,470 --> 00:38:55,410
data is already arranged from largest smallest or
538
00:38:55,410 --> 00:38:58,150
smallest large. It doesn't matter actually. So the
539
00:38:58,150 --> 00:39:02,930
median is $300,000. Make sense? Because there are
540
00:39:02,930 --> 00:39:09,490
two house prices above and two below. So the
541
00:39:09,490 --> 00:39:13,610
median is $300,000. Now if you look at these two
542
00:39:13,610 --> 00:39:21,350
values, the mean for this data equals 600,000 and
543
00:39:21,350 --> 00:39:26,690
the median is 300,000. The mean is double the
544
00:39:26,690 --> 00:39:31,750
median. Do you think why there is a big difference
545
00:39:31,750 --> 00:39:36,030
in this data between the mean and the median?
546
00:39:36,190 --> 00:39:42,290
Which one? Two million dollars is extreme value,
547
00:39:42,510 --> 00:39:45,940
very large number. I mean, if you compare two
548
00:39:45,940 --> 00:39:48,860
million dollars with the other data sets or other
549
00:39:48,860 --> 00:39:51,320
data values, you will see there is a big
550
00:39:51,320 --> 00:39:53,260
difference between two million and five hundred.
551
00:39:53,620 --> 00:39:56,280
It's four times, plus about three hundred
552
00:39:56,280 --> 00:39:59,780
thousands, around seven times and so on. For this
553
00:39:59,780 --> 00:40:07,880
value, the mean is affected. Exactly. The median
554
00:40:07,880 --> 00:40:11,740
is resistant to outliers. It's affected but little
555
00:40:11,740 --> 00:40:17,100
bit. For this reason, we have to use the median.
556
00:40:17,300 --> 00:40:20,720
So the median makes more sense than using the
557
00:40:20,720 --> 00:40:24,480
mean. The mode is just the most frequent value,
558
00:40:24,660 --> 00:40:28,720
which is 100,000, because this value is repeated
559
00:40:28,720 --> 00:40:33,820
twice. So that's the whole story for central
560
00:40:33,820 --> 00:40:40,720
tendency measures, mean, median, and 1. Now the
561
00:40:40,720 --> 00:40:45,640
question again is which measure to use? The mean
562
00:40:45,640 --> 00:40:49,280
is generally used. The most common center tendency
563
00:40:49,280 --> 00:40:53,420
is the mean. We can use it or we should use it
564
00:40:53,420 --> 00:40:59,920
unless extreme values exist. I mean if the data
565
00:40:59,920 --> 00:41:03,960
set has no outliers or extreme values, we have to
566
00:41:03,960 --> 00:41:06,240
use the mean instead of the median.
567
00:41:09,810 --> 00:41:14,670
The median is often used since the median is not
568
00:41:14,670 --> 00:41:18,330
sensitive to extreme values. I mean, the median is
569
00:41:18,330 --> 00:41:22,030
resistant to outliers. It remains nearly in the
570
00:41:22,030 --> 00:41:26,490
same position if the dataset has outliers. But the
571
00:41:26,490 --> 00:41:29,850
median will be affected either to the right or to
572
00:41:29,850 --> 00:41:34,350
the left tail. So we have to use the median if the
573
00:41:34,350 --> 00:41:40,060
data has extreme values. For example, median home
574
00:41:40,060 --> 00:41:44,100
prices for the previous one may be reported for a
575
00:41:44,100 --> 00:41:48,000
region that is less sensitive to outliers. So the
576
00:41:48,000 --> 00:41:52,880
mean is more sensitive to outliers than the
577
00:41:52,880 --> 00:41:56,520
median. Sometimes, I mean in some situations, it
578
00:41:56,520 --> 00:41:58,760
makes sense to report both the mean and the
579
00:41:58,760 --> 00:42:01,860
median. Just say the mean for this data for home
580
00:42:01,860 --> 00:42:07,570
prices is 600,000 while the median is 300,000. If
581
00:42:07,570 --> 00:42:10,150
you look at these two figures, you can tell that
582
00:42:10,150 --> 00:42:13,830
there exists outlier or the outlier exists because
583
00:42:13,830 --> 00:42:17,230
there is a big difference between the mean and the
584
00:42:17,230 --> 00:42:24,310
median. So that's all for measures of central
585
00:42:24,310 --> 00:42:28,830
tendency. Again, we explained three measures,
586
00:42:29,450 --> 00:42:33,930
arithmetic mean, median, and mode. And arithmetic
587
00:42:33,930 --> 00:42:38,990
mean again is denoted by X bar is pronounced as X
588
00:42:38,990 --> 00:42:44,410
bar and just summation of X divided by N. So
589
00:42:44,410 --> 00:42:48,070
summation Xi, i goes from 1 up to N divided by the
590
00:42:48,070 --> 00:42:52,170
total number of observations. The median, as we
591
00:42:52,170 --> 00:42:55,690
mentioned, is the value in the middle in ordered
592
00:42:55,690 --> 00:42:59,150
array. After you arrange the data from smallest to
593
00:42:59,150 --> 00:43:01,930
largest or vice versa, then the median is the
594
00:43:01,930 --> 00:43:06,330
value in the middle. The mode is the most frequent
595
00:43:06,330 --> 00:43:09,030
observed value. And we have to know that mean and
596
00:43:09,030 --> 00:43:13,870
median are used only for numerical data, while the
597
00:43:13,870 --> 00:43:17,510
mode can be used for both numerical and
598
00:43:17,510 --> 00:43:24,290
categorical data. That's all about measures of
599
00:43:24,290 --> 00:43:27,210
central tendency. Any question?
600
00:43:33,210 --> 00:43:40,230
Let's move to measures of variation. It's another
601
00:43:40,230 --> 00:43:43,750
type of measures. It's called measures of
602
00:43:43,750 --> 00:43:47,490
variation, sometimes called measures of spread.
603
00:43:50,490 --> 00:43:53,850
Now, variation can be computed by using range,
604
00:43:55,590 --> 00:44:00,850
variance, standard deviation, and coefficient of
605
00:44:00,850 --> 00:44:08,430
variation. So we have four types, range, variance,
606
00:44:09,250 --> 00:44:12,050
standard deviation, and coefficient of variation.
607
00:44:13,710 --> 00:44:16,150
Now, measures of variation give information on the
608
00:44:16,150 --> 00:44:19,410
spread. Now, this is the first difference between
609
00:44:19,410 --> 00:44:24,210
central tendency measures and measures of
610
00:44:24,210 --> 00:44:28,270
variation. That one measures the central value or
611
00:44:28,270 --> 00:44:30,790
the value in the middle. Here, it measures the
612
00:44:30,790 --> 00:44:36,310
spread. Or variability. Or dispersion of the data.
613
00:44:36,450 --> 00:44:40,310
Do you know what is dispersion? Dispersion.
614
00:44:40,630 --> 00:44:45,590
Tabaad. So major variation given formation with
615
00:44:45,590 --> 00:44:48,350
the spread. Spread or variation or dispersion of
616
00:44:48,350 --> 00:44:52,250
the data values. Now if you look at these two bell
617
00:44:52,250 --> 00:44:52,650
shapes.
618
00:44:55,670 --> 00:44:59,170
Both have the same center. The center I mean the
619
00:44:59,170 --> 00:45:01,730
value in the middle. So the value in the middle
620
00:45:01,730 --> 00:45:06,990
here for figure
621
00:45:06,990 --> 00:45:10,150
graph number one is the same as the value for the
622
00:45:10,150 --> 00:45:16,270
other graph. So both graphs have the same center.
623
00:45:17,430 --> 00:45:20,670
But if you look at the spread, you will see that
624
00:45:20,670 --> 00:45:26,230
figure A is less spread than figure B. Now if you
625
00:45:26,230 --> 00:45:29,720
look at this one, the spread here, is much less
626
00:45:29,720 --> 00:45:34,120
than the other one. Even they have the same
627
00:45:34,120 --> 00:45:39,260
center, the same mean, but figure A is more spread
628
00:45:39,260 --> 00:45:45,140
than figure B. It means that the variation in A is
629
00:45:45,140 --> 00:45:49,920
much less than the variation in figure B. So it
630
00:45:49,920 --> 00:45:55,960
means that the mean is not sufficient to describe
631
00:45:55,960 --> 00:45:59,970
your data. Because maybe you have two datasets and
632
00:45:59,970 --> 00:46:03,330
both have the same mean, but the spread or the
633
00:46:03,330 --> 00:46:07,350
variation is completely different. Again, maybe we
634
00:46:07,350 --> 00:46:10,250
have two classes of statistics, class A and class
635
00:46:10,250 --> 00:46:13,230
B. The center or the mean or the average is the
636
00:46:13,230 --> 00:46:16,150
same for each one. For example, maybe the average
637
00:46:16,150 --> 00:46:19,810
of this class is 70. The average of class B is
638
00:46:19,810 --> 00:46:26,640
also 70. But the scores are scattered. or spread
639
00:46:26,640 --> 00:46:32,580
out in class A maybe much more than in class B. So
640
00:46:32,580 --> 00:46:34,280
the mean is not sufficient to describe the data.
641
00:46:34,360 --> 00:46:37,100
You have to say that the mean equals such and such
642
00:46:37,100 --> 00:46:41,000
and the spread. And one of these measures we'll
643
00:46:41,000 --> 00:46:44,500
talk later about range and variance standard
644
00:46:44,500 --> 00:46:49,030
deviation. So I mean, The mean by itself is not
645
00:46:49,030 --> 00:46:51,890
sufficient to describe the data. You have to use
646
00:46:51,890 --> 00:46:55,730
something else to measure the variation or the
647
00:46:55,730 --> 00:46:57,950
spread of the data. Make sense?
648
00:47:02,170 --> 00:47:05,670
The first measure of variation, the easiest one,
649
00:47:05,810 --> 00:47:11,230
is called the range. The range is the simplest
650
00:47:11,230 --> 00:47:15,590
measure of variation. The range is just the
651
00:47:15,590 --> 00:47:19,750
difference or the distance between the largest and
652
00:47:19,750 --> 00:47:23,550
the smallest value. For example, suppose the
653
00:47:23,550 --> 00:47:27,070
minimum score for this class is 40 and the maximum
654
00:47:27,070 --> 00:47:33,230
is 90. So the range is 50, 90 minus 40. Now
655
00:47:33,230 --> 00:47:38,850
imagine that the minimum score for this class is
656
00:47:38,850 --> 00:47:47,330
60 and the maximum is 80, so 20. If we replace 80
657
00:47:47,330 --> 00:47:51,450
by 100, I mean the minimum is 60 and the maximum
658
00:47:51,450 --> 00:47:57,030
is 100, it's 40. That means a range is affected by
659
00:47:57,030 --> 00:48:02,170
outliers because it depends only on two values.
660
00:48:03,480 --> 00:48:06,100
maximum and minimum value. So it should be
661
00:48:06,100 --> 00:48:09,320
affected by outliers. So range is sensitive to
662
00:48:09,320 --> 00:48:12,780
outliers. So if the data has the data set has
663
00:48:12,780 --> 00:48:15,660
outliers, then in this case, you have to avoid
664
00:48:15,660 --> 00:48:19,640
using range because range only based on two
665
00:48:19,640 --> 00:48:23,480
values. So it should be affected by outliers. Now
666
00:48:23,480 --> 00:48:25,880
for the for simple example, suppose we have this
667
00:48:25,880 --> 00:48:32,360
data. The minimum value is one. I mean, the
668
00:48:32,360 --> 00:48:34,680
smallest value is one, and the largest or the
669
00:48:34,680 --> 00:48:38,880
maximum is 13. So it makes sense that the range of
670
00:48:38,880 --> 00:48:41,840
the data is the difference between these two
671
00:48:41,840 --> 00:48:48,540
values. So 13 minus one is 12. Now, imagine that
672
00:48:48,540 --> 00:48:58,040
we just replace 13 by 100. So the new range will
673
00:48:58,040 --> 00:49:03,820
be equal to 100 minus 199. So the previous range
674
00:49:03,820 --> 00:49:08,340
was 12. It becomes now 99 after we replace the
675
00:49:08,340 --> 00:49:12,100
maximum by 100. So it means that range is affected
676
00:49:12,100 --> 00:49:18,740
by extreme values. So the mean and range both are
677
00:49:18,740 --> 00:49:23,040
sensitive to outliers. So you have to link between
678
00:49:26,410 --> 00:49:30,210
measures of center tendency and measures of
679
00:49:30,210 --> 00:49:33,130
variation. Mean and range are affected by
680
00:49:33,130 --> 00:49:37,910
outliers. The mean and range are affected by
681
00:49:37,910 --> 00:49:41,450
outliers. This is an example. So it's very easy to
682
00:49:41,450 --> 00:49:49,550
compute the mean. Next, if you look at why the
683
00:49:49,550 --> 00:49:51,190
range can be misleading.
684
00:49:53,830 --> 00:49:56,810
Sometimes you report the range and the range does
685
00:49:56,810 --> 00:50:00,310
not give an appropriate answer or appropriate
686
00:50:00,310 --> 00:50:04,450
result because number
687
00:50:04,450 --> 00:50:06,790
one ignores the way in which the data are
688
00:50:06,790 --> 00:50:10,770
distributed. For example, if you look at this
689
00:50:10,770 --> 00:50:15,430
specific data, we have data seven, eight, nine,
690
00:50:15,590 --> 00:50:18,110
ten, eleven and twelve. So the range is five.
691
00:50:19,270 --> 00:50:21,910
Twelve minus seven is five. Now if you look at the
692
00:50:21,910 --> 00:50:26,360
other data, The smallest value was seven.
693
00:50:29,600 --> 00:50:33,260
And there is a gap between the smallest and the
694
00:50:33,260 --> 00:50:38,220
next smallest value, which is 10. And also we have
695
00:50:38,220 --> 00:50:44,480
12 is repeated three times. Still the range is the
696
00:50:44,480 --> 00:50:48,140
same. Even there is a difference between these two
697
00:50:48,140 --> 00:50:53,640
values, between two sets. we have seven, eight,
698
00:50:53,760 --> 00:50:57,020
nine up to 12. And then the other data, we have
699
00:50:57,020 --> 00:51:02,180
seven, 10, 11, and 12 three times. Still, the
700
00:51:02,180 --> 00:51:06,360
range equals five. So it doesn't make sense to
701
00:51:06,360 --> 00:51:09,620
report the range as a measure of variation.
702
00:51:10,520 --> 00:51:12,640
Because if you look at the distribution for this
703
00:51:12,640 --> 00:51:15,500
data, it's completely different from the other
704
00:51:15,500 --> 00:51:20,860
dataset. Even though it has the same range. So
705
00:51:20,860 --> 00:51:25,220
range is not used in this case. Look at another
706
00:51:25,220 --> 00:51:25,680
example.
707
00:51:28,300 --> 00:51:32,920
We have data. All the data ranges, I mean, starts
708
00:51:32,920 --> 00:51:38,680
from 1 up to 5. So the range is 4. If we just
709
00:51:38,680 --> 00:51:46,200
replace the maximum, which is 5, by 120. So the
710
00:51:46,200 --> 00:51:49,190
range is completely different. the range becomes
711
00:51:49,190 --> 00:51:55,010
119. So that means range
712
00:51:55,010 --> 00:51:59,230
is sensitive to outliers. So we have to avoid
713
00:51:59,230 --> 00:52:06,030
using range in case of outliers or extreme values.
714
00:52:08,930 --> 00:52:14,410
I will stop at the most important one, the
715
00:52:14,410 --> 00:52:18,350
variance, for next time inshallah. Up to this
716
00:52:18,350 --> 00:52:19,310
point, any questions?
717
00:52:22,330 --> 00:52:29,730
Okay, stop at this point if
718
00:52:29,730 --> 00:52:30,510
you have any question.
719
00:52:35,430 --> 00:52:39,430
So later we'll discuss measures of variation and
720
00:52:39,430 --> 00:52:44,810
variance, standard deviation up to the end of this
721
00:52:44,810 --> 00:52:45,090
chapter.
722
00:52:54,630 --> 00:53:00,690
So again, the range is sensitive to outliers. So
723
00:53:00,690 --> 00:53:03,850
we have to avoid using range in this case. And
724
00:53:03,850 --> 00:53:06,270
later we'll talk about the variance, which is the
725
00:53:06,270 --> 00:53:09,750
most common measures of variation for next time,
726
00:53:09,830 --> 00:53:10,130
inshallah.