Digital Signal Processing Reference
In-Depth Information
From here, we can easily find the variance as:
2
l
2
1
l
2
¼
1
l
2
:
s
2
¼ m
2
m
1
¼
(2.429)
This is the same result as that obtained in (
2.379
).
2.8.5 Chebyshev Inequality
Sometimes we need to estimate a bound on the probability of how much a random
variable can deviate from its mean value. A mathematical description of this
statement is provided by
Chebyshev inequality
.
The Chebyshev inequality states this bounding in two different forms for r.v.
X
in terms of the expected value
m
X
and the variance
s
X
.
The Chebyshev inequality is very crude, but is useful in situations in which we
have no knowledge of a given random variable, other than its mean value and
variance. However, if we know a density function, then the precise bounds can be
found simply calculating the probability of a desired deviation from the mean value.
2.8.5.1 First Form of Inequality
The probability that the absolute deviation of the random variable
X
from its
expected value
m
X
is more than
e
is less than the variance
s
X
divided by
e
2
,
g s
X
=e
2
PX
f
j
m
X
j e
:
(2.430)
In the continuation, we prove the formula (
2.430
).
Using the definition of the variance, we can write:
1
2
s
X
¼
ðx m
X
Þ
f
X
ðxÞ
d
x
1
ð
ð
1
m
X
e
m
X
þe
2
2
2
¼
ðx m
X
Þ
f
X
ðxÞ
d
x þ
ðx m
X
Þ
f
X
ðxÞ
d
x þ
ðx m
X
Þ
f
X
ðxÞ
d
x:
1
m
X
e
m
X
þe
(2.431)
Omitting the middle integral (see Fig.
2.53a
), we can write:
m
X
e
ð
1
2
2
s
X
ðx m
X
Þ
f
X
ðxÞ
d
x þ
ðx m
X
Þ
f
X
ðxÞ
d
x:
(2.432)
1
m
X
þe
Search WWH ::
Custom Search