Slides for Chapter 2

Order and Rank Statistics
Chapter 2
Order Statistics
โ€ข Assume that a sample of size ๐‘› is selected
from the population.
๐‘‹1 , ๐‘‹2 , โ‹ฏ , ๐‘‹๐‘›
โ€ข The ๐‘˜th order statistic of a statistical sample is
equal to its ๐‘˜th-smallest value.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
2
Order Statistics
โ€ข If ๐‘‹ 1 is the smallest value in the sample, ๐‘‹ 2
is the second smallest value, and so on, then
the order statistics are
๐‘‹1 <๐‘‹2 <โ‹ฏ<๐‘‹๐‘›.
โ€ข ๐‘‹ ๐‘˜ : ๐‘˜th-smallest value in the sample is the
๐‘˜th order statistic of this sample.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
3
Example
โ€ข Assume that a sample of 5 observations is
9, โ€“3, 4, 1, 0
โ€ข Then ๐‘ฅ1 = 9, ๐‘ฅ2 = โˆ’3, ๐‘ฅ3 = 4, ๐‘ฅ4 = 1, ๐‘ฅ5 = 0.
โ€ข The ordered sample is
โ€“3, 0, 1, 4, 9
โ€ข Therefore, the order statistics are
๐‘ฅ 1 = โˆ’3, ๐‘ฅ 2 = 0, ๐‘ฅ 3 = 1, ๐‘ฅ 4 = 4, ๐‘ฅ 5 = 9.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
4
Order Statistics and Their Usage
โ€ข Together with rank statistics, order statistics are
among the most fundamental tools in nonparametric statistics and inference.
โ€ข They are also used to compute some important
statistics:
โ€ข The first order statistic is always the minimum
of the sample, that is,
๐‘‹ 1 = ๐‘š๐‘–๐‘› ๐‘‹1 , ๐‘‹2 , โ‹ฏ , ๐‘‹๐‘› .
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
5
Order Statistics and Their Usage
โ€ข Similarly, for a sample of size ๐‘›, the ๐‘›th order
statistic is the maximum, that is,
๐‘‹ ๐‘› = ๐‘š๐‘Ž๐‘ฅ ๐‘‹1 , ๐‘‹2 , โ‹ฏ , ๐‘‹๐‘› .
โ€ข The range of a sample of size ๐‘›, is the difference
between the ๐‘›th and 1st order statistics, that is,
๐‘… =๐‘‹ ๐‘› โˆ’๐‘‹ 1 .
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
6
Order Statistics and Their Usage
โ€ข The median of a sample is defined in terms of
order statistics:
๐‘‹ ๐‘›+1 ,
if ๐‘› is odd
2
๐‘€= ๐‘‹
๐‘›
2
+๐‘‹
2
๐‘›
+1
2
,
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
if ๐‘› is even
7
Order Statistics and Their Usage
โ€ข The joint (and marginal) distribution of the
order statistics is different from the joint (and
marginal) distribution of the population.
โ€ข Order statistics are dependent.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
8
Example
โ€ข Following is a sample of midterm scores:
๐’Š ๐‘ฟ๐’Š
1 60
๐’Š
๐‘ฟ
๐’Š
2 55
3 80
4 45
1
2
3
4
45
55
60
65
5 70
6 65
7 70
5
6
7
70
70
80
๐‘‹
1
= 45
๐‘‹
4
= 65
๐‘‹
7
= 80
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
๐‘… =๐‘‹ ๐‘› โˆ’๐‘‹ 1
= 80 โˆ’ 45 = 35
9
Rank Order Statistics
โ€ข If ๐‘‹1 , ๐‘‹2 , โ‹ฏ , ๐‘‹๐‘› is a sample of size ๐‘›, then the
rank order statistics are represented by
๐‘Ÿ ๐‘‹1 , ๐‘Ÿ ๐‘‹2 , โ‹ฏ , ๐‘Ÿ ๐‘‹๐‘› .
โ€ข In general, if ๐‘ฅ๐‘– โˆ’ ๐‘ฅ๐‘— = ๐‘ข and
1 ,๐‘ข โ‰ฅ 0
๐‘  ๐‘ข =
0 ,๐‘ข < 0
then
๐‘Ÿ ๐‘‹๐‘– = ๐‘›๐‘—=1 ๐‘  ๐‘ฅ๐‘– โˆ’ ๐‘ฅ๐‘— = 1 + ๐‘›๐‘–โ‰ ๐‘— ๐‘  ๐‘ฅ๐‘– โˆ’ ๐‘ฅ๐‘— .
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
10
Example
โ€ข Assume the following sample is observed:
๐‘ฅ1 = 30, ๐‘ฅ2 = 40, ๐‘ฅ3 = 5, ๐‘ฅ4 = 80, ๐‘ฅ5 = 51, ๐‘ฅ6 = 35.
๐‘Ÿ ๐‘‹1 =
๐‘Ÿ ๐‘‹2 =
๐‘Ÿ ๐‘‹3 =
๐‘Ÿ ๐‘‹4 =
๐‘Ÿ ๐‘‹5 =
๐‘Ÿ ๐‘‹6 =
๐‘›
๐‘—=1 ๐‘ 
๐‘›
๐‘—=1 ๐‘ 
๐‘›
๐‘—=1 ๐‘ 
๐‘›
๐‘—=1 ๐‘ 
๐‘›
๐‘—=1 ๐‘ 
๐‘›
๐‘—=1 ๐‘ 
๐‘ฅ1 โˆ’ ๐‘ฅ๐‘— = 1 + 0 + 1 + 0 + 0 + 0 = 2.
๐‘ฅ2 โˆ’ ๐‘ฅ๐‘— = 1 + 1 + 1 + 0 + 0 + 1 = 4.
๐‘ฅ3 โˆ’ ๐‘ฅ๐‘— = 0 + 0 + 1 + 0 + 0 + 0 = 1.
๐‘ฅ4 โˆ’ ๐‘ฅ๐‘— = 1 + 1 + 1 + 1 + 1 + 1 = 6.
๐‘ฅ5 โˆ’ ๐‘ฅ๐‘— = 1 + 1 + 1 + 0 + 1 + 1 = 5.
๐‘ฅ6 โˆ’ ๐‘ฅ๐‘— = 1 + 0 + 1 + 0 + 0 + 1 = 3.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
11
Example
๐‘ฅ1 = 30, ๐‘ฅ2 = 40, ๐‘ฅ3 = 5,
๐‘ฅ4 = 80, ๐‘ฅ5 = 51, ๐‘ฅ6 = 35
๐‘ฅ 1 = 5, ๐‘ฅ 2 = 30, ๐‘ฅ 3 = 35,
๐‘ฅ 4 = 40, ๐‘ฅ 5 = 51, ๐‘ฅ 6 = 80
๐‘Ÿ ๐‘‹1 = 2, ๐‘Ÿ ๐‘‹2 = 4, ๐‘Ÿ ๐‘‹3 = 1,
๐‘Ÿ ๐‘‹4 = 6, ๐‘Ÿ ๐‘‹5 = 5, ๐‘Ÿ ๐‘‹6 = 3.
Parametrik Olmayan Yöntemler 4. Baskฤฑ
Gamgam, H. & Altunkaynak B
Edited by Güler, H.
12