Sigma 2
Sigma 2
Sigma 2
c 2005
University of Sydney
1
1.1
Sigma Notation
Understanding Sigma Notation
The symbol (capital sigma) is often used as shorthand notation to indicate the sum of a number of similar terms. Sigma notation is used extensively in statistics. For example, suppose we weigh ve children. We will denote their weights by x1 , x2 , x3 , x4 and x5 .
5
xj .
The symbol means add up. Underneath we see j = 1 and on top of it 5. This means that j is replaced by whole numbers starting at the bottom number, 1, until the top number,5, is reached. Thus
5
xj = x2 + x 3 + x4 + x 5 ,
j =2
and
xj = x2 + x3 + x4 .
j =2 n
So the notation
j =1
xj tells us:
a. to add the scores xj , b. where to start: x1 , c. where to stop: xn (where n is some number). Now take the weights of the children to be x1 = 10kg, x2 = 12kg, x3 = 14kg, x4 = 8kg and x5 = 11kg. Then the total weight (in kilograms) is
5
xi = x1 + x2 + x3 + x4 + x5
i=1
= 10 + 12 + 14 + 8 + 11 = 55. Notice that we have used i instead of j in the formula above. The j is what we call a dummy variable - any letter can be used, ie,
n n
xj =
j =1 i=1
xi .
Now let us nd
i=1
Again, starting with i = 1 we replace the expression 2xi with its value and add up the terms until i = 4 is reached. So,
Similarly, let us nd
k=1
(xk 4) where x1 = 7, x2 = 4, x3 = 1.
Here,
3 k=1
xk 4 where x1 = 7, x2 = 4, x3 = 1.
xk 4 = x1 + x2 + x3 4 = 7+4+14 = 8.
We use brackets to indicate what should be included in the sum. In the previous example, there were no brackets, so the 4 was not included in the sum.
5
xk .
j 2 = 12 + 22 + 32 + 42 = 30.
j =1
This is the same principle: replace j in the expression (this time j 2 ) by whole numbers starting with 1 and ending with 4 , and add.
1.1.1
Exercises
4
1. Evaluate
i=1 n
2. Evaluate
k=1
1 5 3. Find = xj where the x1 = 10kg, x2 = 12kg, x3 = 14kg, x4 = 8kg and x5 = 11kg 5 j =1 are the weights of 5 children. ( is the mean weight of the children.)
3
1.2
Rules of summation
We will prove three rules of summation. These rules will allow us to evaluate formulae containing sigma notation more easily and allow us to derive equivalent formulae. Rule 1: If c is a constant, then
n n
cxi = c
i=1 i=1
xi .
To see why Rule 1 is true, lets start with the left hand side of this equation,
n i=1
= c
i=1
xi
as required.
5 5 5
Suppose that
i=1
3xi = 3
i=1
xi = 3 55 = 165.
c = nc.
i=1
This rule looks a bit strange as there is no xi . The left hand side of this formula means sum c, n times. That is,
n i=1 n
c = c + c + + c = nc = nc.
For example,
i=1
2 = 5 2 = 10.
Rule 3:
(xi + yi ) =
i=1 i=1
xi +
i=1
yi .
To prove this rule, lets start with the left hand side,
n i=1
=
i=1 7
xi +
i=1 7
yi .
n 7 7
For example, if
i=1
xi = 21 and
i=1
yi = 35 then
i=1
(xi + yi ) =
i=1
xi +
i=1
yi = 21 + 35 = 56.
1.2.1
Exercises
5 5 5
xi = 37,
i=1
yi = 12,
i=1
x2 i = 303,
xi yi = 105.
1.
i=1 5
2yi xi 1 (xi 1)
5
2.
i=1 5
3.
i=1
4. (
i=1 5
xi )2 (2xi + yi )
5.
i=1 5
6.
i=1 5
7.
i=1 5
8.
i=1
1.3
Here are some examples of how sigma notation is used in statistics: The formula for a mean of a group of N scores, is = 1 N
N
xi .
i=1
A measure of how spread out the scores are, called the variance, has the following formula: 1 = N
2 N i=1
(xi )2 .
For example, the number of customers having lunch at a certain restaurant on 7 weekdays were x1 = 92, x2 = 84, x3 = 70, x4 = 76, x5 = 66, x6 = 80, x7 = 71. The mean is 1 N xi N i=1 1 = (92 + 84 + 70 + 76 + 66 + 80 + 71) 7 539 = 7 = 77.
Note: There are 7 scores so N = 7. The variance is 1 7 (xi 77)2 7 i=1 1 [(15)2 + (7)2 + (7)2 + (1)2 + (11)2 + (3)2 + (6)2 ] 7 1 [225 + 49 + 49 + 1 + 121 + 9 + 36] 7 1 [490] 7 70.
2 = = = = =
For the above example we get: 1 2 2 2 2 2 2 2 [x + x2 2 + x3 + x4 + x5 + x6 + x7 N ] N 1 1 2 [92 + 842 + 702 + 762 + 662 + 802 + 712 7(77)2 ] 7 1 [8464 + 7056 + 4900 + 5776 + 4356 + 6400 + 5041 7(5929)] 7 1 [490] 7 70
2 = = = = =
as before. We can use the rules to show that two formulae for variance are equivalent, since
N i=1
(xi )2 =
N i=1
2 x2 i N .
N i=1
(xi )2 = =
N i=1 N i=1 N
2 (x2 i 2xi + ) N N
x2 i
2xi +
i=1 i=1 N
=
i=1 N
x2 i 2
x i + N 2
i=1
=
i=1 N
2 x2 i 2 N + N
xi = N
=
i=1
2 x2 i N
1.3.1
Exercises
1. Find the variance of the weights of the ve children (in Exercise 1.1.1 number 3), using each of the above formulae for 2 . 2. During a 5 week period, a salespersons weekly income (in dollars) was x1 = 400, x2 = 250, x3 = 175, x4 = 300, x5 = 375. Calculate =
1 5 5 i=1
xi and 2 = 1 ( 5
5 i=1
2 x2 i 5 ).
3. An insurance company is concerned about the length of time required to process claims. The length of time, measured in days, taken to process 7 claims produced the data x1 = 23, x2 = 20, x3 = 22, x4 = 25, x5 = 24, x6 = 23, x7 = 21. Evaluate the mean and variance 2 for these data.
1.4
Answers to Exercises
1.
i=1 5
2yi = 2
i=1
yi = 24
2.
i=1 5
xi 1 = 37 1 = 36 (xi 1) =
5 5 i=1
3.
i=1
xi
5 i=1
1 = 37 5(1) = 32
5
4. (
i=1 5
xi )2 = (37)2 = 1369
5 5
x2 i = 303.
5.
i=1 5
(2xi + yi ) = 2
i=1 5
xi +
i=1
yi = 2(37) + 12 = 86
5
6.
i=1 5
(2xi + 3yi ) = 2
i=1
xi + 3
i=1 5 i=1
7.
i=1 5
xi 5
yi +
i=1 i=1 5
8.
i=1
xi yi + 4
i=1 i=1
i=1
Answers to Exercise 1.3.1 1. 2 = 4 2. = 300, 2 = 6750 3. = 22.57 to two decimal places, 2 = 2.53 taking the mean as 22.57 and using the formula 2 =
1 N (x N i=1 i
)2 .
1 2 2 2 If, however the formula 2 = N [N i=1 xi N ] is used, then the answer = 2.60 is obtained. This discrepancy is due to round o error and can be avoided by using = 22.571429 in the above formula.