Professional Documents
Culture Documents
Appendix A
A Compendium of Common Probability Distributions
Version 2.3
PREFACE
This Appendix contains summaries of the probability distributions found in Regress+.
All distributions are shown in their parameterized, not standard forms. In some cases, the
definition of a distribution may vary slightly from a definition given in the literature. This
happens either because there is more than one definition or, in the case of parameters,
because Regress+ requires a parameter to be constrained, usually to guarantee convergence.
There are a few well-known distributions that are not included here, either because they
are seldom used to model empirical data or they lack a convenient analytical form for the
CDF. Conversely, many of the distributions that are included are rarely discussed yet are
very useful for describing real-world datasets.
Please email any comments to the author:
mpmcl@mitre.org
Michael P. McLaughlin
McLean, VA
September, 1999
A-3
A-4
Table of Contents
Distributions shown in bold are those which appear in the Regress+ menu. Distributions
shown in plain text are aliases and/or special cases.
Continuous Distributions
Name
Antilognormal
Bell curve
Beta(A,B,C,D)
Bilateral exponential
Bradford(A,B,C)
Burr(A,B,C,D)
Cauchy(A,B)
Chi(A,B,C)
Chi-square
Cobb-Douglas
Cosine(A,B)
Double-exponential
DoubleGamma(A,B,C)
DoubleWeibull(A,B,C)
Erlang
Error function
Exponential(A,B)
Extreme-value
ExtremeLB(A,B,C)
Fisher-Tippett
Fisk(A,B,C)
FoldedNormal(A,B)
Frechet
Gamma(A,B,C)
Gaussian
GenLogistic(A,B,C)
Gompertz
Gumbel(A,B)
Page
77
85
9
63
15
17
19
21
41
77
23
63
25
27
41
85
29
119
35
49
37
39
119
41
85
43
49
49
Name
HalfNormal(A,B)
HyperbolicSecant(A,B)
Inverse Gaussian
InverseNormal(A,B)
Laplace(A,B)
Logistic(A,B)
LogLogistic
LogNormal(A,B)
LogWeibull
Lorentz
Maxwell
Negative exponential
Nakagami(A,B,C)
Non-central Chi
Normal(A,B)
Pareto(A,B)
Power-function
Rayleigh
Reciprocal(A,B)
Rectangular
Sech-squared
Semicircular(A,B)
StudentsT(A,B,C)
Triangular(A,B,C)
Uniform(A,B)
Wald
Weibull(A,B,C)
A-5
Page
51
59
61
61
63
75
37
77
49
19
21
29
79
39
85
99
9
21
105
113
75
107
109
111
113
61
119
Continuous Mixtures
Name
Double double-exponential
Expo(A,B)&Expo(A,C)
Expo(A,B)&Uniform(A,C)
HNormal(A,B)&Expo(A,C)
HNormal(A,B)&HNormal(A,C)
HNormal(A,B)&Uniform(A,C)
Laplace(A,B)&Laplace(C,D)
Laplace(A,B)&Laplace(A,C)
Laplace(A,B)&Uniform(C,D)
Laplace(A,B)&Uniform(A,C)
Normal(A,B)&Laplace(C,D)
Normal(A,B)&Laplace(A,C)
Normal(A,B)&Normal(C,D)
Normal(A,B)&Normal(A,C)
Normal(A,B)&Uniform(C,D)
Normal(A,B)&Uniform(A,C)
Uniform(A,B)&Uniform(C,D)
Uniform(A,B)&Uniform(A,C)
Schuhl
Page
65
31
33
53
55
57
65
67
69
71
87
89
91
93
95
97
115
117
31
Discrete Distributions
Name
Binomial(A,B)
Furry
Geometric(A)
Logarithmic(A)
NegativeBinomial(A,B)
Pascal
Poisson(A)
Polya
Page
11
45
45
73
81
81
101
81
Discrete Mixtures
Name
Binomial(A,C)&Binomial(B,C)
Geometric(A)&Geometric(B)
NegBinomial(A,C)&NegBinomial(B,C)
Poisson(A)&Poisson(B)
A-6
Page
13
47
83
103
A-7
Legend
Shown below are definitions for some of the less common functions and abbreviations
used in this Appendix.
int(y)
(z)
erf(z)
error function
(z)
(w,x)
I(x,y,z)
H(n)
(s)
N
k
EulerGamma = 0.57721566...
i.i.d.
iff
if and only if
N!
(is) distributed as
A-8
Beta(A,B,C,D)
3.0
H0, 1, 6, 2L
2.5
H0, 1, 1, 2L
2.0
H0, 1, 2, 2L
1.5
1.0
0.5
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Y
PDF =
C+D
C D BA
CDF = I
C+D1
yA
C1
By
D1
yA
, C, D
BA
CD BA
C+D+1 C+D
2CD DC
Skewness =
C+D
C+D+1 C+D+2
A-9
CD
2
C+D C+D+1
3
2
C2 D + 2 + 2 D2 + C D D 2
Kurtosis = 3
Mode =
C+D+1
CD C+D+2 C+D+3
A D1 +B C1
, unless C = D = 1
C+D2
A-10
y = 0, 1, 2, , 0 < A < 1, y, 2 B
Binomial(A,B)
0.25
H0.45, 10L
0.20
0.15
0.10
0.05
0.00
0
Y
PDF =
B
Ay 1 A
y
By
Mode = int A B + 1
A-11
10
Notes
1. In the literature, B may be any positive integer.
2. If A (B + 1) is an integer, Mode also equals A (B + 1) 1.
3. Regress+ requires B to be Constant.
Aliases and Special Cases
1. Although disallowed here because of incompatibility with several Regress+ features,
Binomial(A,1) is called the Bernoulli distribution.
Characterizations
1. The probability of exactly y successes, each having Prob(success) = A, in a series of B
independent trials is ~Binomial(A, B).
A-12
Binomial(A,C)&Binomial(B,C)
y = 0, 1, 2, , 0 < B < A < 1, y, 3 C, 0 < p < 1
0.25
0.20
H0.6, 0.2, 10, 0.25L
0.15
0.10
0.05
0.00
0
10
Y
PDF =
C
y
p Ay 1 A
Cy
+ 1 p By 1 B
Cy
Variance = C p A 1 A + 1 p p C A B + B 1 B
A-13
Notes
1. Here, parameter A is stipulated to be the Component with the larger Prob(success).
2. Parameter C must be at least 3 in order for this distribution to be identifiable,
i.e., well-defined.
3. Regress+ requires C to be Constant.
4. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
5. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-14
Bradford(A,B,C)
3.0
H0, 1, 5L
2.0
H0, 1, 1L
1.0
0.0
0.0
0.5
Y
C
PDF =
C y A + B A log C + 1
log 1 +
CDF =
C yA
BA
log C + 1
Ck
BA
Variance =
C k2 +2k
2 C k2
A-15
1.0
2 12 C 2 9 k C C + 2 + 2 k 2 C C + 3 + 3
Skewness =
C C k2 +2k
3C k2 +6k
C 3 k 3 k 3 k 16 + 24 + 12 k C 2 k 4 k 3 + 6 C k 2 3 k 14 + 12 k 3
Kurtosis =
3C C k2 +2k
Mode = A
Median = 1 A C + 1 B + B A
C
Q1 = 1 A C + 1 B + B A
C
C+1
log
qMean =
C+1
Q3 = 1 A C + 1 B + B A
C
C+1
C
log C + 1
log C + 1
qMode = 0
RandVar = 1 A C + 1 B + B A C + 1
C
Notes
1. With the log-likelihood criterion, parameter C is often flat.
Aliases and Special Cases
Characterizations
1. The Bradford distribution has been used to model the distribution of references among
several sources.
A-16
Burr(A,B,C,D)
0.80
H0, 1, 2, 1L
0.60
H0, 2, 3, 2L
0.40
0.20
0.00
0
Y
yA
PDF = C D
B
B
C1
yA
1+
B
yA
CDF = 1 +
B
D1
B 1 1 1 +D
C
C
Mean = A +
D
2
Variance = k2 B
D
A-17
Skewness
3 D
k3
2 3 1 1 3 1 + D
3 1 2 1 1 1 +D 2 +D
1 3 3 +D
C
C
C
C
C
C
C
C
+
3 D
2 D
D
Kurtosis =
4 D
3+
k2
4 D
k2
6 1 2 2 1 1 2 1 + D 2 + D
3 4 1 1 4 1 + D
C
C
C
C
C
C
+
4
3
D
D
1 4 4 +D
4 1 3 1 1 1 +D 3 +D
C
C
C
C
C
C
2
D
D
Mode = A + B
Q1 = A + B
4 1
1
C
2 1
1
C
Q3 = A + B
4 1
3
1
C
qMean = 1 +
1 1
C
1 +D
C
RandVar = A + B D1 1
u
qMode = 1 + C + 1
CD1
1
C
Notes
1. With the log-likelihood criterion, parameter C is often flat.
Aliases and Special Cases
1. The Burr distribution, with D = 1, is often called the Fisk or LogLogistic distribution.
Characterizations
1. The Burr distribution is a generalization of the Fisk distribution.
A-18
Cauchy(A,B)
B>0
0.6
0.5
0.4
0.3
H3, 0.6L
0.2
H0, 1L
0.1
0.0
-8
-6
-4
-2
Y
1
PDF =
B 1+
yA
B
1 tan 1 y A
CDF = 1 +
B
2
Q3 = A + B
qMode = 0.5
A-19
RandVar = A + B tan u 1
2
Notes
1. Since there are no finite moments, the location parameter (ostensibly the mean) does not
have its usual interpretation for a symmetrical distribution.
Aliases and Special Cases
1. The Cauchy distribution is sometimes called the Lorentz distribution.
Characterizations
1. If U and V are ~Normal(0, 1), the ratio U/V ~Cauchy(0, 1).
2. If Z ~Cauchy, then W = (a + b*Z)-1 ~Cauchy.
3. If particles emanate from a fixed point, their points of impact on a straight line ~Cauchy.
A-20
Chi(A,B,C)
H0, 1, 1L
0.8
H0, 1, 2L
0.6
0.4
H0, 1, 3L
0.2
0.0
0
PDF =
yA
B
C1
yA
exp 1
B
2
C
22 1 B C
2
yA
CDF = C , 1
B
2 2
A-21
2 4 3 C + 1 + 2 C
2
2
2 C+3 3C C+1
2
2
Skewness =
2 C+1
2
C
2 C
2
2
3 C
2
Kurtosis =
3
2
2 C 1 C 4 C 24 4 C + 1 + 8 2 C 1 2 C 2 C + 1
2
2
2
2
2
C 2 2 C + 1
2
2
Mode = A + B C 1
A-22
A B y A + B, B > 0
Cosine(A,B)
0.4
0.3
H0, 1L
0.2
0.1
0.0
-3
-2
-1
Y
PDF =
1 1 + cos y A
B
2B
yA
yA
CDF = 1 +
+ sin
B
B
2
Skewness = 0
Kurtosis =
6 4 90
5 2 6
Q1 A 0.8317 B
0.5938
Q3 A + 0.8317 B
A-23
Notes
Aliases and Special Cases
Characterizations
1. The Cosine distribution is sometimes used as a simple, and more computationally
tractable, approximation to the Normal distribution.
A-24
DoubleGamma(A,B,C)
B, C > 0
H0, 1, 3L
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
-10
-5
Y
PDF =
1
2B C
CDF =
yA
B
C1
exp
yA
B
1 1 C, y A
B
2 2
, yA
1 + 1 C, y A
B
2 2
, y>A
A-25
10
Notes
Aliases and Special Cases
Characterizations
1. The DoubleGamma distribution is the signed version of the Gamma distribution.
A-26
DoubleWeibull(A,B,C)
B, C > 0
0.6
H0, 1, 3L
0.5
0.4
0.3
0.2
0.1
0.0
-2
-1
Y
yA
PDF = C
B
2B
CDF =
C1
yA
B
exp
1 exp y A
B
2
yA
1 1 exp
B
2
yA
,
C
, y>A
A-27
Notes
Aliases and Special Cases
Characterizations
1. The DoubleWeibull distribution is the signed version of the Weibull distribution.
A-28
y A, B > 0
Exponential(A,B)
1.0
H0, 1L
0.8
0.6
0.4
H0, 2L
0.2
0.0
0
Y
Ay
PDF = 1 exp
B
B
CDF = 1 exp
Ay
B
Variance = B 2
Skewness = 2
Kurtosis = 6
Mode = A
Median = A + B log 2
A-29
Q1 = A + B log 4
3
Q3 = A + B log 4
qMean = e e 1 0.6321
qMode = 0
RandVar = A B log u
Notes
1. The one-parameter version of this distribution, Exponential(0,B), is far more common
than the more general formulation shown here.
Aliases and Special Cases
1. The Exponential distribution is sometimes called the Negative exponential distribution.
2. The discrete version of the Exponential distribution is the Geometric distribution.
Characterizations
1. If the future lifetime of a system at any time, t, has the same distribution for all t, then this
distribution is the Exponential distribution. This is known as the memoryless property.
A-30
Expo(A,B)&Expo(A,C)
1.0
0.8
H0, 1, 2, 0.7L
0.6
0.4
0.2
0.0
0
Y
PDF =
1p
p
Ay
Ay
exp
+
exp
B
B
C
C
CDF = p stdExponentialCDF
yA
yA
+ 1 p stdExponentialCDF
B
C
Variance = C 2 + 2 B B C p B C p 2
Mode = A
A-31
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
1. This mixture, when applied to traffic analysis, is often called the Schuhl distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-32
Expo(A,B)&Uniform(A,C)
1.0
0.8
H0, 1, 2, 0.7L
0.6
0.4
0.2
0.0
0
Y
PDF =
1p y<C
p
Ay
exp
+
B
B
CA
yA
yA
+ 1p
, y<C
B
CA
yA
p stdExponentialCDF
+ 1p , yC
B
p stdExponentialCDF
CDF =
A + C + p (A + 2 B C)
2
2
2
Variance = p B 2 + A + B
p1 A +AC+C
3
Mode = A
A-33
A+C+p A+2BC
A-34
ExtremeLB(A,B,C)
0.9
H1, 1, 2L
0.8
0.7
0.6
0.5
0.4
0.3
H1, 2, 3L
0.2
0.1
0.0
1
Y
yA
PDF = C
B
B
C1
exp
CDF = exp
yA
B
yA
B
Skewness =
C 3 3 C 2 C 1 + 2 3 C 1
C
C
C
C
3
C 2 2 C 1
C
C
A-35
Kurtosis = 6 +
C 4 4 C 3 C 1 + 3 2 C 2
C
C
C
C
2
C 2 2 C 1
C
C
C
Mode = A + B
Median = A +
Q1 = A +
B
log 4
C
1+C
B
log 2
B
Q3 = A +
C
qMean = exp C C 1
C
log 4
3
qMode = exp C + 1
C
RandVar = A + B log u
1
C
Notes
1. The name ExtremeLB does not appear in the literature. It was chosen here simply to
indicate one type of extreme-value distribution with a lower bound.
2. In the literature, C > 0. The restriction shown above is required for convergence when
the data are left-skewed.
3. Moment k exists if C > k.
Aliases and Special Cases
1. The corresponding distribution with an upper bound is Weibull(y).
Characterizations
1. Extreme-value distributions are the limiting distributions, as N --> infinity, of the
greatest value among N i.i.d. variates selected from a continuous distribution. By
replacing y with y, the smallest values may be modeled.
A-36
Fisk(A,B,C)
0.7
H0, 1, 2L
0.6
0.5
0.4
0.3
H0, 2, 3L
0.2
0.1
0.0
0
PDF = C
B
CDF =
yA
B
C1
yA
1+
B
1
yA
1+
B
Mean = A + B csc
C
C
A-37
Skewness =
3
2
Kurtosis =
3 3 csc 4 12 C 2 csc csc 3 + 4 C 3 csc 4 + 6 C 2 csc 3 sec
C
C
C
C
C
C
csc 2 2 C csc 2
C
C
Mode = A + B
C1
C+1
Median = A + B
Q1 = A + CB
3
qMean =
Q3 = A + B C 3
1+
csc
C
C
RandVar = A + B C
qMode = C 1
2C
u
1u
Notes
1. The Fisk distribution is right-skewed.
2. To model a left-skewed distribution, try modeling w = y.
3. Moment k exists if C > k.
Aliases and Special Cases
1. The Fisk distribution is also known as the LogLogistic distribution.
Characterizations
1. The Fisk distribution is often used in income and lifetime analysis.
A-38
y 0, A 0, B > 0
FoldedNormal(A,B)
H1, 1L
0.5
0.4
0.3
0.2
H1, 2L
0.1
0.0
0
Y
PDF = 1
B
2
2
2 cosh A y exp 1 y + A
2 B2
B2
CDF =
yA
yA
B
B
Parameters -- A (): Location, B (): Scale, both for the corresponding unfolded Normal
Moments, etc.
Mean = B
2
1 A
exp 2 B
A 12 A
B
2
Variance = A 2 + B 2 B
2 exp A 2 + A erf
2 B2
A-39
A
2B
3 A2
2
exp 2 B 2
Skewness =
2 A erf
A
2B
2
4 B 2 exp A 2
B
2 A2 + B2
+
Var 3
2
2
6 B 2 exp A 2 + 3 2 A B exp A 2 erf
B
2B
A + A 2 erf 2
2B
A 1
2B
Var 3
Kurtosis = 3 +
2
A4 + 6 A2 B2 + 3 B4 + 6 A2
+ B2
A2
2
exp 2 B 2 + A erf
A
2B
Var 2
4
3 B
Var 2
A2
2
exp 2 B 2 + A erf
2
4 exp A 2
A
2B
A2
2
+ A exp 2 B 2 erf
A
2B
A 2 + 2 B 2 + A A 2 + 3 B 2 exp
A 2 erf
2 B2
A
2B
Var 2
A-40
Gamma(A,B,C)
1.0
0.8
H0, 1, 1L
0.6
H0, 1, 2L
0.4
0.2
0.0
0
Y
1
PDF =
B C
yA
B
CDF = C,
C1
exp
Ay
B
yA
B
Variance = B 2 C
Skewness = 2
C
Kurtosis = 6
C
Mode = A + B C 1
qMean = C, C
qMode = C, C 1
Notes
1. The Gamma distribution is right-skewed.
2. To model a left-skewed distribution, try modeling w = y.
3. The Gamma distribution approaches a Normal distribution in the limit as C goes to
infinity.
4. In the literature, C > 0. The restriction shown above is required primarily to recognize
when the PDF is not right-skewed.
Aliases and Special Cases
1. Gamma(A, B, C), where C is an integer, is the Erlang distribution.
2. Gamma(A, B, 1) is the Exponential distribution.
3. Gamma(0, 2, /2) is the Chi-square distribution with degrees of freedom.
Characterizations
1. If Z1 ~Gamma(A, B, C1) and Z2 ~Gamma(A, B, C2), then (Z1 + Z2)
~Gamma(A, B, C1 + C2).
Z k ~Gamma(0, 2, /2).
2. If Z1, Z2, , Z ~Normal(0, 1), then W = k
=1
2
Z k ~Erlang(A, B, n).
3. If Z1, Z2, , Zn ~Exponential(A, B), then W = k
=1
A-42
GenLogistic(A,B,C)
B, C > 0
0.4
0.3
H0, 1, 2L
0.2
0.1
0.0
-4
-2
PDF = C
B
exp
Ay
B
C+1
Ay
1 + exp
B
1
CDF =
Ay
1 + exp
B
Skewness =
C + 2 3
2 + C
6
A-43
3
2
12 4 + 15 C
Kurtosis =
5 2 + 6 C
Mode = A + B log C
Median = A B log
Q1 = A B log
4 1
2 1
Q3 = A B log
4 1
3
qMean = 1 + exp H C 1
RandVar = A B log
qMode =
C
C+1
1 1
u
Notes
1. The Generalized Logistic distribution is a generalization of the Logistic distribution.
2. The Generalized Logistic distribution is left-skewed when C < 1 and right-skewed for
C > 1.
3. There are additional generalizations of the Logistic distribution.
Aliases and Special Cases
1. The Generalized Logistic distribution becomes the Logistic distribution when C = 1.
Characterizations
1. The Generalized Logistic has been used in the analysis of extreme values.
A-44
y = 1, 2, 3, , 0 < A < 1
Geometric(A)
H0.45L
0.5
0.4
0.3
0.2
0.1
0.0
0
Y
PDF = A 1 A
y1
Mode = 1
A-45
10
Notes
Aliases and Special Cases
1. The Geometric distribution is the discrete version of the Exponential distribution.
2. The Geometric distribution is sometimes called the Furry distribution.
Characterizations
1. In a series of Bernoulli trials, with Prob(success) = A, the number of trials required to
realize the first success is ~Geometric(A).
2. For the Bth success, see the NegativeBinomial distribution.
A-46
Geometric(A)&Geometric(B)
0.20
0.15
0.10
0.05
0.00
0
10
12
14
16
Y
PDF = p A 1 A
y1
+ 1p B 1B
y1
Variance =
p 1p
+
B
A
A2 1 B + p B A B A 2 p2 A B
A2 B2
Mode = 1
A-47
18
20
Notes
1. Here, parameter A is stipulated to be the Component with the larger Prob(success).
2. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
3. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-48
Gumbel(A,B)
B>0
0.40
0.35
H0, 1L
0.30
0.25
0.20
0.15
0.10
H1, 2L
0.05
0.00
-4
-2
Y
Ay
Ay
PDF = 1 exp
exp exp
B
B
B
Ay
B
Mean = A + B
Variance = 1 B
6
Skewness =
12 6 3
1.1395
3
Kurtosis = 12
5
Mode = A
A-49
10
Q1 = A B log log 4
Q3 = A B log log 4
3
0.5704
qMode = 1e 0.3679
Notes
1. The Gumbel distribution is one of the class of extreme-value distributions.
2. The Gumbel distribution is right-skewed.
3. To model a left-skewed distribution, try modeling w = y.
Aliases and Special Cases
1. The Gumbel distribution is sometimes called the LogWeibull distribution.
2. It is also known as the Gompertz distribution.
3. It is also known as the Fisher-Tippett distribution.
Characterizations
1. Extreme-value distributions are the limiting distributions, as N --> infinity, of the greatest
value among N i.i.d. variates selected from a continuous distribution. By replacing y
with y, the smallest values may be modeled.
2. The Gumbel distribution is often used to model maxima when the random variable is
unbounded.
A-50
y A, B > 0
HalfNormal(A,B)
0.8
H0, 1L
0.7
0.6
0.5
0.4
0.3
H0, 2L
0.2
0.1
0.0
0
Y
PDF = 1
B
2
1 yA
exp 2
B
CDF = 2
yA
1
B
Mean = A + B
2
Variance = B 2 1
Skewness =
Kurtosis =
2 4
2
8 3
2
A-51
0.9953
0.8692
Mode = A
Median A + 0.6745 B
Q1 A + 0.3186 B
Q3 A + 1.150 B
qMean 0.5751
qMode = 0
Notes
Aliases and Special Cases
1. The HalfNormal distribution is a special case of both the Chi and the FoldedNormal
distributions.
Characterizations
1. If X ~Normal(A, B) is folded (to the right) about its mean, A, the resulting distribution is
HalfNormal(A, B).
A-52
HNormal(A,B)&Expo(A,C)
0.8
0.7
H0, 1, 2, 0.7L
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
Y
PDF =
2 p
1 yA
B exp 2
B
CDF = p stdHalfNormalCDF
1p
Ay
exp
C
C
yA
yA
+ 1 p stdExponentialCDF
B
C
2
+ 1p C
1 p 2 p B2 + 2 p p 1 B C 2 p2 1 C2
Variance =
Mode = A
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. The alternate, Expo(A,B)&HNormal(A,C) distribution may be obtained by switching
identities in the parameter dialog. In this case, the parameters shown above in the
Moments section must be reversed (cf. E&E and H&H).
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-54
HNormal(A,B)&HNormal(A,C)
0.7
H0, 1, 2, 0.7L
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
PDF =
2 p
1 yA
B exp 2
B
CDF = p stdHalfNormalCDF
1p
yA
exp 1
2
C
C
yA
yA
+ 1 p stdHalfNormalCDF
B
C
2
pB+ 1p C
2 pB+ 1p C
Variance = p B 2 + 1 p C 2
Mode = A
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-56
HNormal(A,B)&Uniform(A,C)
0.8
0.7
H0, 1, 2, 0.7L
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
Y
PDF =
2 p
1 yA
B exp 2
B
1p y<C
CA
yA
yA
+ 1p
, y<C
B
CA
yA
p stdHalfNormalCDF
+ 1p , yC
B
p stdHalfNormalCDF
CDF =
A-57
2 + AC
1p 1+3p
A-58
HyperbolicSecant(A,B)
B>0
0.6
0.5
0.4
0.3
H3, 0.6L
0.2
H0, 1L
0.1
0.0
-8
-6
-4
-2
Y
sech
PDF =
yA
B
B
2 tan 1 exp y A
CDF =
B
Skewness = 0
Kurtosis = 2
Q1 = A B log 1 + 2
Q3 = A + B log 1 + 2
A-59
Notes
1. The Hyperbolic Secant is related to the Logistic distribution.
Aliases and Special Cases
Characterizations
1. If Z ~Hyperbolic Secant, then W = exp(Z) ~Half Cauchy.
2. The Hyperbolic Secant distribution is used in lifetime analysis.
A-60
InverseNormal(A,B)
y > 0, A, B > 0
1.2
H1, 1L
1.0
0.8
0.6
0.4
H2, 3L
0.2
0.0
0
Y
PDF =
CDF =
B exp B y A
A
2y
2 y3
B y A + exp 2 B
y A
A
B yA
y
A
Skewness = 3
A
B
Kurtosis = 15 A
B
Mode = A
2B
9 A2 + 4 B2 3 A
A-61
A-62
Laplace(A,B)
B>0
1.0
0.8
0.6
H3, 0.6L
0.4
H0, 1L
0.2
0.0
-6
-4
-2
PDF =
CDF =
1 exp y A
B
2B
1 exp y A ,
yA
B
2
Ay
1 1 exp
, y>A
B
2
Q3 = A + B log 2
A-63
Notes
Aliases and Special Cases
1. The Laplace distribution is often called the double-exponential distribution.
2. It is also known as the bilateral exponential distribution.
Characterizations
1. The Laplace distribution is the signed analogue of the Exponential distribution.
2. Errors of real-valued observations are often ~Laplace or ~Normal.
A-64
Laplace(A,B)&Laplace(C,D)
0.4
0.3
0.1
0.0
-4
-2
PDF =
yA
p
exp
B
2B
CDF = p stdLaplaceCDF
yC
1p
exp
D
2D
yA
yC
+ 1 p stdLaplaceCDF
B
D
Variance = p 2 B 2 p 1 A C
2 p 1 D2
A-65
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
1. This binary mixture is very often referred to as the Double double-exponential
distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-66
Laplace(A,B)&Laplace(A,C)
0.5
0.4
H0, 1, 2, 0.7L
0.3
0.2
0.1
0.0
-6
-4
-2
PDF =
yA
p
exp
B
2B
CDF = p stdLaplaceCDF
yA
1p
exp
2C
C
yA
yA
+ 1 p stdLaplaceCDF
B
C
Variance = 2 p B 2 + 1 p C 2
Skewness = 0
6 p B4 + 1 p C4
Kurtosis =
p B2 + 1 p C2
RandVar: determined by p
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
1. This is a special case of the Laplace(A,B)&Laplace(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-68
Laplace(A,B)&Uniform(C,D)
0.5
0.4
H0, 1, -1, 1, 0.9L
0.3
0.2
0.1
0.0
-5
-4
-3
-2
-1
PDF =
yA
p
exp
B
2B
1p C<y<D
DC
yA
, yC
B
yA
yC
p stdLaplaceCDF
+ 1p
, C<y<D
B
DC
yA
p stdLaplaceCDF
+ 1p , yD
B
p stdLaplaceCDF
CDF =
1p C+D
2
2
1p C+D
Variance = p A 2 + 2 B 2 p A +
2
A-69
+ 1 1 p C2 + C D + D2
3
A-70
Laplace(A,B)&Uniform(A,C)
0.6
0.5
H0, 1, 1, 0.7L
0.4
0.3
0.2
0.1
0.0
-4
-3
-2
-1
Y
yA
p
PDF =
exp
B
2B
1p
+
AC <y< A+C
2C
yA
, yAC
B
yA
yA+C
p stdLaplaceCDF
+ 1p
, AC<y<A+C
B
2C
yA
p stdLaplaceCDF
+ 1p , yA+C
B
p stdLaplaceCDF
CDF =
A-71
Skewness = 0
9 120 p B 4 + 1 p C 4
Kurtosis =
2
5 6pB + 1p C
RandVar: determined by p
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. Note that the parameters for the Uniform component are different from those in the
Uniform(A,B) distribution. Here, A is the Mean and C is the half-width.
Aliases and Special Cases
1. This is a special case of the Laplace(A,B)&Uniform(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-72
y = 1, 2, 3, , 0 < A < 1
Logarithmic(A)
H0.45L
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0
Y
PDF =
Ay
log 1 A y
A
1 A log 1 A
A A + log 1 A
Variance =
1 A log 2 1 A
Mode = 1
A-73
Notes
Aliases and Special Cases
Characterizations
1. The Logarithmic distribution has been used to model the numbers of items of a product
purchased by a buyer in a given period of time.
A-74
Logistic(A,B)
B>0
0.5
0.4
0.3
H3, 0.6L
0.2
H0, 1L
0.1
0.0
-8
-6
-4
-2
PDF = 1
B
yA
B
exp
yA
1 + exp
B
1
CDF =
1 + exp
Ay
B
A-75
Q1 = A B log 3
Q3 = A + B log 3
u
1u
Notes
1. The logistic distribution is often used as an approximation to other symmetrical
distributions due to the mathematical tractability of its CDF.
Aliases and Special Cases
1. The logistic distribution is sometimes called the Sech-squared distribution.
Characterizations
1. The logistic law of growth is described by the following differential equation:
dP = P P 2
dt
whence
Pt =
P0 P
P0 + P P0 exp t
A-76
LogNormal(A,B)
y > 0, B > 0
1.0
H0.5, 2L
0.8
0.6
0.4
H0, 1L
0.2
0.0
0
Y
log y A
1
PDF =
exp 1
B
2
By 2
CDF =
log y A
B
A-77
Median = exp A
Q1 exp A 0.6745 B
Q3 exp A + 0.6745 B
A-78
Nakagami(A,B,C)
1.6
H0, 1, 1L
1.4
1.2
1.0
0.8
H0, 2, 3L
0.6
0.4
0.2
0.0
0
Y
PDF =
2 CC
B C
yA
B
2C1
exp C
yA
CDF = C, C
B
yA
B
A-79
B2
2 3 C + 1 + 2 C C + 3 3 C C + 1
2
2
2
Skewness =
3
2
C 2 C + 1
2
C C 1
2
C+1
3
Kurtosis =
6 4 C + 1 3 C2 4 C + 3 C C + 2 + 23 4 C 4 C 1 2 2 C
2
2
C + 1 C 2 C
2
2
Mode = A + B
2C1
2C
A-80
y = 1, 2, 3, , 0 < A < 1, 1 B y
NegativeBinomial(A,B)
0.12
H0.45, 5L
0.10
0.08
0.06
0.04
0.02
0.00
5
10
15
20
Y
PDF =
y1
AB 1 A
B1
yB
B 1A
A2
Mode = int A + B 1
A
A-81
25
Notes
1. Although not supported here, the NegativeBinomial distribution may be generalized to
include non-integer values of B.
2. If (B 1)/A is an integer, Mode also equals (B 1)/A .
3. Regress+ requires B to be Constant.
Aliases and Special Cases
1. The NegativeBinomial is also known as the Pascal distribution. The latter name is
restricted to the case (as here) in which B is an integer.
2. It is also known as the Polya distribution.
3. If B = 1, the NegativeBinomial distribution becomes the Geometric distribution.
Characterizations
1. If Prob(success) = A, the number of Bernoulli trials required to realize the Bth success is
~NegativeBinomial(A, B).
A-82
NegBinomial(A,C)&NegBinomial(B,C)
y = 1, 2, 3, , 0 < B < A < 1, 1 C y, 0 < p < 1
0.010
0.005
0.000
10
15
20
25
30
35
40
45
50
Y
PDF =
y1
C1
p AC 1 A
yC
+ 1 p BC 1 B
yC
1p
p
+
B
A
C p B2 1 + C 1 p A2 1 p B p C 1 p A B B + 2 C 1 p
Variance =
A2 B2
A-83
Notes
1. Here, parameter A is stipulated to be the Component with the larger Prob(success).
2. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
3. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-84
Normal(A,B)
B>0
0.7
0.6
H3, 0.6L
0.5
0.4
H0, 1L
0.3
0.2
0.1
0.0
-4
-2
Y
PDF =
1 exp 1 y A
B
2
B 2
CDF =
yA
B
Variance = B 2
Skewness = Kurtosis = 0
Q1 A 0.6745 B
Q3 A + 0.6745 B
A-85
Notes
1. The CDF is generally tabulated in terms of the standard variable z:
z =
yA
B
N s
N1
Characterizations
1. Let Z1, Z2, , ZN be i.i.d. random variables with finite values for their mean () and
variance (2). Then, for any real number (z),
lim Prob
N
1
N
i=1
Zi
z = z
known as the Central Limit Theorem. Loosely speaking, the sum of k random variables,
from the same distribution, tends to be ~Normal, and more so as k increases.
2. If X ~Normal(A, B), then Y = a X + b ~Normal(a A + b, a B).
3. If X ~Normal(A, B) and Y ~Normal(C, D), then S = X + Y
2
2
(i.e., the convolution of X and Y) is ~ Normal A + C, B + D
A-86
Normal(A,B)&Laplace(C,D)
0.3
H0, 1, 3, 0.6, 0.7L
0.2
0.1
0.0
-4
-2
PDF =
p
yA
exp 1
B
2
B 2
CDF = p
yC
1p
exp
D
2D
yA
yC
+ 1 p stdLaplaceCDF
B
D
Variance = p B 2 p 1 A C
2 p 1 D2
A-87
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. The alternate, Laplace(A,B)&Normal(C,D) distribution may be obtained by switching
identities in the parameter dialog. In this case, the parameters shown above in the
Moments section must be reversed (cf. L&L and N&N).
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-88
Normal(A,B)&Laplace(A,C)
0.4
H0, 1, 2, 0.7L
0.3
0.2
0.1
0.0
-5
-4
-3
-2
-1
PDF =
p
yA
exp 1
B
2
B 2
CDF = p
yA
1p
exp
2C
C
yA
yA
+ 1 p stdLaplaceCDF
B
C
Variance = p B 2 + 2 1 p C 2
Skewness = 0
Kurtosis =
3 p B 4 + 24 1 p C 4
p B2 + 2 1 p C2
RandVar: determined by p
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. The alternate, Laplace(A,B)&Normal(A,C) distribution may be obtained by switching
identities in the parameter dialog. In this case, the parameters shown above in the
Moments section must be reversed (cf. L&L and N&N).
Aliases and Special Cases
1. This is a special case of the Normal(A,B)&Laplace(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-90
Normal(A,B)&Normal(C,D)
0.3
H0, 1, 3, 0.6, 0.7L
0.2
0.1
0.0
-4
-2
Y
PDF =
yA
p
exp 1
B
2
B 2
CDF = p
yC
1p
exp 1
D
2
D 2
yA
yC
+ 1p
B
D
Variance = p B 2 p 1 A C
p 1 D2
A-91
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. Whether or not this much-studied mixture is bimodal depends partly upon parameter p.
Obviously, if p is small enough, this mixture will be unimodal regardless of the
remaining parameters. If
2
2
2
A C > 8 2B D 2
B +D
then there will be some values of p for which this mixture is bimodal. However, if
AC
<
27 B 2 D 2
4 B2 + D2
A-92
Normal(A,B)&Normal(A,C)
0.4
H0, 1, 2, 0.7L
0.3
0.2
0.1
0.0
-6
-4
-2
Y
PDF =
yA
p
exp 1
B
2
B 2
CDF = p
yA
1p
exp 1
C
2
C 2
yA
yA
+ 1p
B
C
Variance = p B 2 + 1 p C 2
Skewness = 0
Kurtosis =
3 p 1 p B2 C2
p B2 + 1 p C2
RandVar: determined by p
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
1. This is a special case of the Normal(A,B)&Normal(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-94
Normal(A,B)&Uniform(C,D)
0.5
0.4
H0, 1, -1, 1, 0.9L
0.3
0.2
0.1
0.0
-4
-3
-2
-1
Y
p
yA
PDF =
exp 1
B
2
B 2
1p C<y<D
DC
yA
, yC
B
yA
yC
p
+ 1p
, C<y<D
B
DC
yA
p
+ 1p , yD
B
p
CDF =
1p C+D
2
2
1p C+D
Variance = p A 2 + B 2 p A +
2
A-95
+ 1 1 p C2 + C D + D2
3
A-96
Normal(A,B)&Uniform(A,C)
0.5
H0, 1, 1, 0.7L
0.4
0.3
0.2
0.1
0.0
-3
-2
-1
Y
p
yA
PDF =
exp 1
B
2
B 2
1p
AC <y< A+C
2C
yA
, yAC
B
yA
yA+C
p
+ 1p
, AC<y<A+C
B
2C
yA
p
+ 1p , yA+C
B
p
CDF =
A-97
Skewness = 0
9 15 p B 4 + 1 p C 4
Kurtosis =
2
5 3pB + 1p C
RandVar: determined by p
Notes
1. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
2. Warning! Mixtures usually have several local optima.
3. It is especially difficult to get the optimum value for parameter C in this distribution.
There is an unusual amount of ambiguity in this case. It might be best to start out with
parameter C Constant at its likely value (if known) and proceed from there.
Aliases and Special Cases
1. This is a special case of the Normal(A,B)&Uniform(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-98
0 < A y, B > 0
Pareto(A,B)
2.00
H1, 2L
1.75
1.50
1.25
1.00
0.75
0.50
H1, 1L
0.25
0.00
1
Y
B
PDF = BBA+ 1
y
CDF = 1 A
y
Variance =
Skewness =
A2 B
(B 2) (B 1) 2
2 B+1
B3
B2
B
6 B3 + B2 6 B 2
Kurtosis =
B B 2 7 B + 12
A-99
Mode = A
Median = A B 2
Q1 = A
4
3
qMean = 1 B 1
B
Q3 = A B 4
B
qMode = 0
RandVar = BA
u
Notes
1. The Pareto distribution is always right-skewed.
2. There are several alternate forms for this distribution. In fact, the term Pareto is often
applied to a class of distributions.
3. Moment k exists if B > k.
Aliases and Special Cases
Characterizations
1. The Pareto distribution is often used as an income distribution. Thus, the probability that
a random income, in some defined population, exceeds a minimum, A, is ~Pareto.
A-100
y = 0, 1, 2, , A > 0
Poisson(A)
0.25
H3.5L
0.20
0.15
0.10
0.05
0.00
0
Y
PDF =
exp A A y
y!
A-101
10
12
Notes
1. If A is an integer, Mode also equals A 1.
Aliases and Special Cases
Characterizations
1. The Poisson distribution is commonly used as an approximation to the Binomial(A)
distribution when A is very small.
2. In queueing theory, when interarrival times are ~Exponential, the number of arrivals in a
fixed interval are ~Poisson.
3. Errors in observations with integer values (i.e., miscounting) are ~Poisson.
A-102
Poisson(A)&Poisson(B)
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0
10
12
14
16
Y
PDF =
p exp A A y + 1 p exp B B y
y!
Variance = p A A + 1 + 1 p B B + 1 p A B + B
A-103
18
20
Notes
1. Here, parameter A is stipulated to be the Component with the smaller expectation.
2. This distribution may or may not be bimodal.
3. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
4. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-104
0<AyB
Reciprocal(A,B)
0.25
0.20
0.15
0.10
H1, 100L
0.05
0.00
0
20
40
60
Y
PDF =
1
y log B
A
log A
y
CDF =
log A
B
Parameters -- A, B: Shape
Moments, etc. [d log (A/B)]
Mean = A B
d
AB A d2 +B d+2
Variance =
2 d2
A-105
80
100
2 12 d A B + d 2 A 2 2 d 9 + 2 A B d + B 2 2 d + 9
Skewness =
3
2
3d AB A d2 +B d+2
3
Kurtosis =
36 A B + 36 d A B
A + B 16 d 2 A 3 B 3 + 3 d 3 A 2 + B 2 A + B
2
3 AB A d2 +B d+2
Mode = A
Median =
Q1 =
A3 4 B
AB
Q3 =
qMean = 1 log A d
AB
d
B3
qMode = 0
RandVar = A 1 u B u
Notes
1. The Reciprocal distribution is unusual in that it has no location or scale parameter.
Aliases and Special Cases
Characterizations
1. The Reciprocal distribution is often used to describe 1/f noise.
A-106
A B y A + B, B > 0
Semicircular(A,B)
H0, 1L
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-1
Y
PDF =
2
B
1 yA
CDF = 1 +
B
2
yA
B
yA
B
+ asin
yA
B
Skewness = 0
Kurtosis = 1
Q1 A 0.4040 B
Q3 A + 0.4040 B
A-107
Notes
Aliases and Special Cases
Characterizations
1. The Semicircular distribution, like the Cosine distribution, is sometimes used as an
alternative to the Normal distribution.
A-108
StudentsT(A,B,C)
0.8
H3, 0.5, 10L
0.7
0.6
0.5
0.4
H0, 1, 2L
0.3
0.2
0.1
0.0
-5
-4
-3
-2
-1
PDF =
C+1
2
B C C
2
yA
B
1+
C
C+1
2
yA
0
B
C , C, 1 , t y A > 0
1 1 I
B
2 C + t2 2 2
C , C, 1 ,
1I
2 C + t2 2 2
CDF =
C B2
C2
Skewness = 0
A-109
2
C2 C 2
2
Kurtosis = 3
1
4 C
2
Notes
1. In the literature, C > 0. The bounds shown here are used to prevent divergence.
2. The Students-t distribution approaches the Normal distribution asymptotically as
C -> infinity.
3. If the optimum model has C = 100, use Normal instead.
4. Moment k exists if C > k.
Aliases and Special Cases
1. The Students-t distribution is often referred to as simply the t-distribution.
Characterizations
1. The Students-t distribution is used to characterize small samples (typically, N < 30) from
a Normal population.
A-110
Triangular(A,B,C)
A < y, C < B
0.3
0.2
H-4, 4, 2L
0.1
0.0
-4
-3
-2
-1
Y
2 yA
PDF =
BA CA
2 By
BA BC
yA
CDF =
, y<C
, yC
BA CA
, y<C
A B C + B C 2 y + y2
AB BC
, yC
Mean = A + B + C
3
A-111
Variance = d
18
2 A+B2C 2ABC A2B+C
Skewness =
5 d3
Kurtosis = 3
5
Mode = C
Median = A + 1
2
B A C A , if C A + B else B 1
2
2
Q1 = A + 1
2
B A C A , if qMode 1 else B 1
4
2
Q3 = A + 1
2
3 B A C A , if qMode 3 else B 1
4
2
B+C2A
qMean =
BA BC
3 BA BC
BA BC
9 BA CA
, C A+B
2
A2 + 5 A B 5 B2 7 A C + 5 B C + C2 , C < A + B
2
9 AB BC
qMode = C A
BA
RandVar = A +
u B A C A , if qMode u else B
1u BA BC
Notes
Aliases and Special Cases
Characterizations
1. If X is ~Uniform(a,b) and Z is ~Uniform(c,d) and (b a) = (d c), then (X + Z) is
~Triangular(a+c,b+d,(a+b+c+d)/2). The latter is called the convolution of the two
Uniform distributions.
A-112
Uniform(A,B)
5
A<y<B
H-1.6, -1.4L
2
H-0.5, 0.5L
-2
-1
H1, 3L
Y
PDF =
1
BA
CDF =
yA
BA
Variance = 1 B A
12
Skewness = 0
Kurtosis = 6
5
Mode = none
Q1 = 3 A + B
4
Q3 = A + 3 B
4
A-113
qMean = 0.5
RandVar = A + u B A
Notes
1. The parameters for the Uniform distribution are sometimes given, equivalently, as the
mean and half-width of the domain.
Aliases and Special Cases
1. The Uniform distribution is often called the Rectangular distribution.
Characterizations
1. The Uniform distribution is commonly used to describe total ignorance within a bounded
interval.
2. Pseudo-random number generators typically return X ~Uniform(0, 1) as their primary
output.
A-114
Uniform(A,B)&Uniform(C,D)
0.3
H-3, 2, 1, 3, 0.7L
0.2
0.1
0.0
-3
-2
-1
Y
PDF =
1p C<y<D
p y<B
+
BA
DC
CDF = cdf1 + cdf2
where
cdf1 = p if y > B , else
p yA
and
BA
1p yC
DC
p A+B + 1p C+D
2
p A2 + A B + B2 3 p A + B + 1 p C + D
4
Variance =
3
+ 1 p C2 + C D + D2
RandVar: determined by p
Notes
1. As implemented here, this distribution requires overlap between the two components.
Also, Component #1 must cover min(y) although either Component may cover max(y).
2. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
3. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-116
Uniform(A,B)&Uniform(A,C)
0.5
H0, 1, 2, 0.7L
0.4
0.3
0.2
0.1
0.0
-2
-1
Y
p AB <y< A+B
PDF =
CDF =
2B
1p
2C
1p
yA+C , yAB
2C
p
1p
yA+B +
yA+C , AB<y<A+B
2B
2C
1p
p+
yA+C , yA+B
2C
Variance =
p B2 + 1 p C2
3
Skewness = 0
A-117
9 p B4 + 1 p C4
Kurtosis =
2
5 pB + 1p C
Mode = none
RandVar: determined by p
Notes
1. Note that the parameters are defined differently here than in all other Uniform
distributions discussed elsewhere in this document.
2. Binary mixtures may require hundreds of data points for adequate optimization and, even
then, often have unusually wide confidence intervals. In fact, the criterion response is
sometimes flat over a broad range, esp. with respect to parameter p.
3. Warning! Mixtures usually have several local optima.
Aliases and Special Cases
1. This is a special case of the Uniform(A,B)&Uniform(C,D) distribution.
Characterizations
1. The usual interpretation of a binary mixture is that it represents an undifferentiated
composite of two populations having parameters and weights as described above.
A-118
Weibull(A,B,C)
y > A, B, C > 0
1.0
H1, 1, 1L
0.8
0.6
H1, 2, 3L
0.4
0.2
0.0
1
Y
yA
PDF = C
B
B
C1
exp
CDF = 1 exp
yA
B
yA
B
Skewness =
2 3 C + 1 3 C + 1 C + 2 + C + 3
C
C
C
C
3
C + 2 2 C + 1
C
C
A-119
Kurtosis =
3 4 C + 1 + 6 2 C + 1 C + 2 4 C + 1 C + 3 + C + 4
C
C
C
C
C
C
2
A, C1
Mode =
A+B
C1 , C>1
C
Median = A + B
Q1 = A + B
C+1 C+2
C
C
log 4
3
qMean = 1 exp C C + 1
C
log 2
Q3 = A + B
log 4
RandVar = A + B C log u
Notes
1. The Weibull distribution is roughly symmetrical for C near 3.6. When C is
smaller/larger, the distribution is left/right-skewed, respectively.
Aliases and Special Cases
1. The Weibull is sometimes known as the Frechet distribution.
2. Weibull(y) is the upper-bound analogue of the ExtremeLB distribution, one of a class
of Extreme-value distributions.
3. Weibull(A,B,1) is the Exponential distribution.
4. Weibull(0,1,2) is the standard Rayleigh distribution.
Characterizations
C
yA
is ~ standard Exponential, then y ~Weibull(A, B, C). Thus, the
B
Weibull distribution is a generalization of the Exponential distribution.
2. Weibull(y) is an extreme-value distribution for variates with an upper bound.
1. If X =
A-120