Seoul National University Instructor: Junho Song

Dept. of Civil and Environmental Engineering [email protected]

457.212 Statistics for Civil & Environmental Engineers

In-Class Material: Class 11
Useful Distribution Models – Part I (A&T: 3.2)

“Distribution models” are useful because ...

• The probability function is the result of an underlying physical process and can be
derived on the basis of certain physically reasonable assumptions.
• The function is the result of some limiting process (e.g., ).
• The distribution is widely known, and the necessary probability and statistical
information (including probability tables) are widely available
(e.g., probability table of )

1. Normal distribution

• Best known and most widely used. Also known as _______________ distribution.
• According to ________________________, the sum of random variables converges to
a normal random variable as the number of the variables increases, no matter what
distributions the variables are subjected to.
• Completely defined by the ______________ and the ____________________ of the
random variable.

(a) PDF: 𝑋~𝑁(μ, σ2 )

1  1  x −  2 
f X ( x) = exp −   , −  x  
2   2    

(b) CDF: no closed-form expression available

x
FX ( x ) = f X ( x)dx , −   x  
−
(c) Parameters: μ, σ

• μ: _________ of the random variable, i.e. μ=μ X  E[ X ]

σ: ____________________ of the random variable, i.e. σ = σ X  E[( X − μ X )2 ]
0.5
•

(d) Shape of the PDF plots

• Symmetric around x =
• A change in μ X _______________ the PDF horizontally by the same amount.
• The larger the value of σ X gets, the more ______________ the PDF becomes
around the central axis.
Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

(e) R functions

dnorm(0, mean=0, sd=1) # PDF of normal distribution

pnorm(0, mean=0, sd=1) # CDF of normal distribution
qnorm(0.3, mean=0, sd=1) # inverse CDF of normal distribution
rnorm(5, mean=0, sd=1)
# generates random numbers from normal distribution
x=seq(-5,5,0.1)
plot(x,dnorm(x),type="l")
plot(x,pnorm(x),type="l")
xr = rnorm(100000)
hist(xr,freq=FALSE,breaks=seq(-5,5,0.1))

0.08

0.07

0.06
 = 0,  = 10  = 20,  = 10 =40,  = 10
0.05
f (x)

0.04
X

0.03

0.02

0.01

0
-20 -10 0 10 20 30 40 50 60
x

Figure 1. PDF’s of normal random variables with different values of 

0.08

 = 20,  = 5
0.07

0.06

0.05
f (x)

0.04
X

 = 20,  = 10
0.03

0.02  = 20,  = 15

0.01

0
-20 -10 0 10 20 30 40 50 60
x

Figure 2. PDF’s of normal random variables with different values of 

Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

1a. Standard normal distribution

• A special case of the normal distribution: μ X = , σX = .

• The CDF of the standard normal distribution can be used for computing the CDF of any
general normal random variable.

(a) PDF: U ~ N( , )
1  1 
(u ) = exp − u 2  , −   u  
2  2 

(b) CDF:
u
 (u ) =  (u )du , − u  
−

→ no closed-form expression available, but the table of the standard normal CDF ()
can be found in books or computer software (e.g., See Appendix A of A&T)

(c) Inverse CDF of standard normal distribution:  −1 ()

(u p ) = p  u p =  −1 ( p )

(d) Symmetry around u = :

(−u ) = 1 − (u )
u1− p = −u p

→ The table of the standard normal CDF is often provided for positive u values only,
but using the symmetry one can find the CDF for negative values as well.

(e) One can compute the CDF of a general normal random variable 𝑋~𝑁(μ, σ2 ) by use of
the CDF of the standard normal random variable 𝑈~𝑁(0,12 ) as follows.

FX (a ) = P( X  a )
a
1  1  x − μ 2 
= 
− 2 σ
exp  −    dx
 2  σ  
 a −μ 
  1  1 
=  σ 
exp  − u 2  σdu
−
2 σ  2 
 
=  
 
   
Hence, P (a  X  b) = FX ( ) − FX ( ) =  − 
   
Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

Example 1: Given a standard normal distribution, find the area under the curve that lies

(a) to the right of u = 1.84

(b) between u = −1.97 and u = 0.86

1-pnorm(1.84, mean=0, sd=1) # (a)

pnorm(0.86, mean=0, sd=1) - pnorm(-1.97, mean=0, sd=1) # (b)

Example 2: The drainage demand during a storm (in mgd: million gallons/day):
𝑋~𝑁(1.2,0.42 ). The maximum drain capacity is 1.5 mgd.

(a) Probability of flooding?

(b) Probability that the drainage demand during a storm will be between 1.0 and 1.6 mgd?

(c) The 90-percentile drainage demand?

1 - pnorm(1.5, mean=1.2, sd=0.4) # (a)

pnorm(1.6, mean=1.2, sd=0.4) - pnorm(1.0, mean=1.2, sd=0.4) # (b)
qnorm(0.9, mean=1.2, sd=0.4) # (c)
Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

2. Lognormal distribution

• Closely related to the ________ distribution.

• Defined for _________ values only.

(a) PDF: 𝑋~𝐿𝑁(λ, ζ2 )

1  1  ln x −   2 
f X ( x) = exp −    , 0  x  
2  x  2    

(b) CDF:
x
FX ( x ) = f
−
X ( x)dx , 0  x  

→ no closed-form expression available, but can be computed by use of the table of the
standard normal CDF  () (as shown below)

(c) Parameters: λ, ζ

• λ: mean of _____________, i.e. λ = λ X  E[ln X ]

• ζ: standard deviation of______________, i.e. ζ 2 = ζ 2X = σln2 X

(d) Shape of the PDF plots

1.8 1
 = 0,  = 0.25  = 0,  = 0.5
0.9  = 1,  = 0.5
1.6  = 0,  = 0.5
 = 0,  = 1  = 2,  = 0.5
0.8
1.4

0.7
1.2
Lognormal PDF

Lognormal PDF

0.6
1
0.5
0.8
0.4
0.6
0.3

0.4
0.2

0.2 0.1

0 0
0 0.5 1 1.5 2 2.5 3 0 1 2 3 4 5 6 7 8 9 10
x x

Figure 3. PDF’s of lognormal random variables.

(e) R functions

dlnorm(1, meanlog=0, sdlog=1) # PDF of lognormal dist.

plnorm(1, meanlog=0, sdlog=1) # CDF of lognormal dist.
qlnorm(0.3, meanlog=0, sdlog=1) # inverse CDF of lognormal dist.
rlnorm(5, meanlog=0, sdlog=1)
# generates random numbers from lognormal dist.
Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

(f) Relationship between normal and lognormal distribution:

“The logarithm of a ___________ random variable is a __________ random variable.”

𝑋~𝐿𝑁(λ, ζ2 )  ln𝑋~𝑁(μ = λ, σ2 = 𝜁 2 )

(g) Can obtain the CDF of lognormal 𝑋~𝐿𝑁(λ, ζ2 ) from the CDF of standard normal:

FX (𝑎) = 𝑃(𝑋 ≤ 𝑎)
= 𝑃(ln𝑋 ≤ ln𝑎) since ln𝑋~𝑁(μ = λ, σ2 = 𝜁 2 )
ln𝑎 − 𝜆
= Φ( )
𝜁

(h) “The exponential function of a _____________ random variable is a ______________

random variable.”

𝑋~𝐿𝑁(λX , ζ2X ) 𝑌~𝑁(μ𝑌 = λ𝑥 , σ2𝑌 = 𝜁𝑋2 )

(i) (, ) → (, ) : Find the mean and c.o.v. from the distribution parameters

 = E[ X ] = exp(  + 0.5 2 )
 =  /  = exp( 2 ) − 1 (  for   1)

(j) (, ) → (, ) : Find the distribution parameters from the mean and c.o.v.

 = ln(1 +  2 ) (  for   1)
 = ln − 0.5 ln(1 +  2 )

(k) ( x0.5 )  ( ) : Relationship between the median and 

 = ln x0.5 , x0.5 = e 

(l) (, ) → ( x0.5 ) : Find the median from the mean and c.o.v.

x0.5 =
1 + 2

Note: x 0.5   for the lognormal distribution.

Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

Example 3: The drainage demand during a storm (in mgd: million gallons/day) is assumed
to follow the lognormal distribution with the same mean and standard deviation as Example
1 (mean 1.2, standard deviation 0.4). The maximum drain capacity is 1.5 mgd.

(a) Distribution parameters, i.e. λ and ζ?

(b) Probability of the flooding?

(c) Probability that the drainage demand during a storm will be between 1.0 and 1.6 mgd?

(d) The 90-percentile drainage demand?

# (a)
cov = 0.4/1.2
lambda = log(1.2)-0.5*log(1+cov^2)
zeta = sqrt(log(1+cov^2))

# (b)
1 - plnorm(1.5, meanlog=lambda, sdlog=zeta)

# (c)
Prob_16 = plnorm(1.6, meanlog=lambda, sdlog=zeta)
Prob_10 = plnorm(1.0, meanlog=lambda, sdlog=zeta)
Prob_16 – Prob_10

# (d)
qlnorm(0.9, meanlog=lambda, sdlog=zeta)
Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

Example 4: Consider a bridge whose uncertain capacity against “complete damage” limit-
state caused by earthquake events is defined in terms of peak ground acceleration (PGA;
unit: g) that the bridge can sustain. Suppose the median of the capacity is 1.03g and the
coefficient of variation is 0.50. It is assumed that the capacity follows a lognormal
distribution.

(a) Distribution parameters of the lognormal distribution, i.e. λ and ζ?

(b) The mean and standard deviation of the uncertain capacity, i.e.  and ?

(c) Suppose the peak ground acceleration from an earthquake event is 0.5g. What is the
probability that the structure will exceed “complete damage” limit state?
Seoul National
x University x
PHI(x) PHI(x) x PHI(x) x PHI(x) Instructor:
x Junho
PHI(x) Song
Dept. of
0.00Civil and0.5Environmental
0.90 Engineering 1.80
0.81593987 0.96406968 2.70 0.99653303 [email protected]
3.60 0.999840891
0.01 0.50398936 0.91 0.81858875 1.81 0.96485211 2.71 0.99663584 3.61 0.999846901
0.02 0.50797831 0.92 0.82121362 1.82 0.9656205 2.72 0.9967359 3.62 0.999852698
0.03 0.51196647 0.93 0.82381446 1.83 0.96637503 2.73 0.99683328 3.63 0.999858289
0.04 0.51595344 0.94 0.82639122 1.84 0.96711588 2.74 0.99692804 3.64 0.999863681
0.05 0.51993881 0.95 0.82894387 1.85 0.96784323 2.75 0.99702024 3.65 0.99986888
0.06 0.52392218 0.96 0.83147239 1.86 0.96855724 2.76 0.99710993 3.66 0.999873892
0.07 0.52790317 0.97 0.83397675 1.87 0.96925809 2.77 0.99719719 3.67 0.999878725
0.08 0.53188137 0.98 0.83645694 1.88 0.96994596 2.78 0.99728206 3.68 0.999883383
0.09 0.53585639 0.99 0.83891294 1.89 0.97062102 2.79 0.9973646 3.69 0.999887873
0.10 0.53982784 1.00 0.84134475 1.90 0.97128344 2.80 0.99744487 3.70 0.9998922
0.11 0.54379531 1.01 0.84375235 1.91 0.97193339 2.81 0.99752293 3.71 0.99989637
0.12 0.54775843 1.02 0.84613577 1.92 0.97257105 2.82 0.99759882 3.72 0.999900389
0.13 0.55171679 1.03 0.848495 1.93 0.97319658 2.83 0.9976726 3.73 0.99990426
0.14 0.55567 1.04 0.85083005 1.94 0.97381016 2.84 0.99774432 3.74 0.99990799
0.15 0.55961769 1.05 0.85314094 1.95 0.97441194 2.85 0.99781404 3.75 0.999911583
0.16 0.56355946 1.06 0.8554277 1.96 0.9750021 2.86 0.99788179 3.76 0.999915043
0.17 0.56749493 1.07 0.85769035 1.97 0.97558081 2.87 0.99794764 3.77 0.999918376
0.18 0.57142372 1.08 0.85992891 1.98 0.97614824 2.88 0.99801162 3.78 0.999921586
0.19 0.57534543 1.09 0.86214343 1.99 0.97670453 2.89 0.99807379 3.79 0.999924676
0.20 0.57925971 1.10 0.86433394 2.00 0.97724987 2.90 0.99813419 3.80 0.999927652
0.21 0.58316616 1.11 0.86650049 2.01 0.97778441 2.91 0.99819286 3.81 0.999930517
0.22 0.58706442 1.12 0.86864312 2.02 0.97830831 2.92 0.99824984 3.82 0.999933274
0.23 0.59095412 1.13 0.87076189 2.03 0.97882173 2.93 0.99830519 3.83 0.999935928
0.24 0.59483487 1.14 0.87285685 2.04 0.97932484 2.94 0.99835894 3.84 0.999938483
0.25 0.59870633 1.15 0.87492806 2.05 0.97981778 2.95 0.99841113 3.85 0.999940941
0.26 0.60256811 1.16 0.8769756 2.06 0.98030073 2.96 0.9984618 3.86 0.999943306
0.27 0.60641987 1.17 0.87899952 2.07 0.98077383 2.97 0.998511 3.87 0.999945582
0.28 0.61026125 1.18 0.88099989 2.08 0.98123723 2.98 0.99855876 3.88 0.999947772
0.29 0.61409188 1.19 0.8829768 2.09 0.9816911 2.99 0.99860511 3.89 0.999949878
0.30 0.61791142 1.20 0.88493033 2.10 0.98213558 3.00 0.9986501 3.90 0.999951904
0.31 0.62171952 1.21 0.88686055 2.11 0.98257082 3.01 0.99869376 3.91 0.999953852
0.32 0.62551583 1.22 0.88876756 2.12 0.98299698 3.02 0.99873613 3.92 0.999955726
0.33 0.62930002 1.23 0.89065145 2.13 0.98341419 3.03 0.99877723 3.93 0.999957527
0.34 0.63307174 1.24 0.8925123 2.14 0.98382262 3.04 0.99881711 3.94 0.999959259
0.35 0.63683065 1.25 0.89435023 2.15 0.98422239 3.05 0.99885579 3.95 0.999960924
0.36 0.64057643 1.26 0.89616532 2.16 0.98461367 3.06 0.99889332 3.96 0.999962525
0.37 0.64430875 1.27 0.89795768 2.17 0.98499658 3.07 0.99892971 3.97 0.999964064
0.38 0.64802729 1.28 0.89972743 2.18 0.98537127 3.08 0.998965 3.98 0.999965542
0.39 0.65173173 1.29 0.90147467 2.19 0.98573788 3.09 0.99899922 3.99 0.999966963
0.40 0.65542174 1.30 0.90319952 2.20 0.98609655 3.10 0.9990324 4.00* 3.16712E-05
0.41 0.65909703 1.31 0.90490208 2.21 0.98644742 3.11 0.99906456 4.05 2.56088E-05
0.42 0.66275727 1.32 0.90658249 2.22 0.98679062 3.12 0.99909574 4.10 2.06575E-05
0.43 0.66640218 1.33 0.90824086 2.23 0.98712628 3.13 0.99912597 4.15 1.66238E-05
0.44 0.67003145 1.34 0.90987733 2.24 0.98745454 3.14 0.99915526 4.20 1.33457E-05
0.45 0.67364478 1.35 0.91149201 2.25 0.98777553 3.15 0.99918365 4.25 1.06885E-05
0.46 0.67724189 1.36 0.91308504 2.26 0.98808937 3.16 0.99921115 4.30 8.53991E-06
0.47 0.68082249 1.37 0.91465655 2.27 0.98839621 3.17 0.99923781 4.35 6.80688E-06
0.48 0.6843863 1.38 0.91620668 2.28 0.98869616 3.18 0.99926362 4.40 5.41254E-06
0.49 0.68793305 1.39 0.91773556 2.29 0.98898934 3.19 0.99928864 4.45 4.29351E-06
0.50 0.69146246 1.40 0.91924334 2.30 0.98927589 3.20 0.99931286 4.50 3.39767E-06
0.51 0.69497427 1.41 0.92073016 2.31 0.98955592 3.21 0.99933633 4.55 2.68230E-06
0.52 0.69846821 1.42 0.92219616 2.32 0.98982956 3.22 0.99935905 4.60 2.11245E-06
0.53 0.70194403 1.43 0.92364149 2.33 0.99009692 3.23 0.99938105 4.65 1.65968E-06
0.54 0.70540148 1.44 0.9250663 2.34 0.99035813 3.24 0.99940235 4.70 1.30081E-06
0.55 0.70884031 1.45 0.92647074 2.35 0.99061329 3.25 0.99942297 4.75 1.01708E-06
0.56 0.71226028 1.46 0.92785496 2.36 0.99086253 3.26 0.99944294 4.80 7.93328E-07
0.57 0.71566115 1.47 0.92921912 2.37 0.99110596 3.27 0.99946226 4.85 6.17307E-07
0.58 0.71904269 1.48 0.93056338 2.38 0.99134368 3.28 0.99948096 4.90 4.79183E-07
0.59 0.72240468 1.49 0.93188788 2.39 0.99157581 3.29 0.99949906 4.95 3.71068E-07
0.60 0.72574688 1.50 0.9331928 2.40 0.99180246 3.30 0.99951658 5.00 2.86652E-07
0.61 0.7290691 1.51 0.93447829 2.41 0.99202374 3.31 0.99953352 5.10 1.69827E-07
0.62 0.73237111 1.52 0.93574451 2.42 0.99223975 3.32 0.99954991 5.20 9.96443E-08
0.63 0.73565271 1.53 0.93699164 2.43 0.99245059 3.33 0.99956577 5.30 5.79013E-08
0.64 0.7389137 1.54 0.93821982 2.44 0.99265637 3.34 0.99958111 5.40 3.33204E-08
0.65 0.74215389 1.55 0.93942924 2.45 0.99285719 3.35 0.99959594 5.50 1.89896E-08
0.66 0.74537309 1.56 0.94062006 2.46 0.99305315 3.36 0.99961029 5.60 1.07176E-08
0.67 0.7485711 1.57 0.94179244 2.47 0.99324435 3.37 0.99962416 5.70 5.99037E-09
0.68 0.75174777 1.58 0.94294657 2.48 0.99343088 3.38 0.99963757 5.80 3.31575E-09
0.69 0.75490291 1.59 0.9440826 2.49 0.99361285 3.39 0.99965054 5.90 1.81751E-09
0.70 0.75803635 1.60 0.94520071 2.50 0.99379033 3.40 0.99966307 6.00 9.86588E-10
0.71 0.76114793 1.61 0.94630107 2.51 0.99396344 3.41 0.99967519 6.10 5.30342E-10
0.72 0.7642375 1.62 0.94738386 2.52 0.99413226 3.42 0.99968689 6.20 2.82316E-10
0.73 0.76730491 1.63 0.94844925 2.53 0.99429687 3.43 0.99969821 6.30 1.48823E-10
0.74 0.77035 1.64 0.94949742 2.54 0.99445738 3.44 0.99970914 6.40 7.76885E-11
0.75 0.77337265 1.65 0.95052853 2.55 0.99461385 3.45 0.99971971 6.50 4.01600E-11
0.76 0.77637271 1.66 0.95154277 2.56 0.99476639 3.46 0.99972991 6.60 2.05579E-11
0.77 0.77935005 1.67 0.95254032 2.57 0.99491507 3.47 0.99973977 6.70 1.04210E-11
0.78 0.78230456 1.68 0.95352134 2.58 0.99505998 3.48 0.99974929 6.80 5.23093E-12
0.79 0.78523612 1.69 0.95448602 2.59 0.9952012 3.49 0.99975849 6.90 2.60014E-12
0.80 0.7881446 1.70 0.95543454 2.60 0.99533881 3.50 0.99976737 7.00 1.27987E-12
0.81 0.79102991 1.71 0.95636706 2.61 0.99547289 3.51 0.99977595 7.10 6.23834E-13
0.82 0.79389195 1.72 0.95728378 2.62 0.99560351 3.52 0.99978423 7.20 3.01092E-13
0.83 0.79673061 1.73 0.95818486 2.63 0.99573076 3.53 0.99979222 7.30 1.43885E-13
0.84 0.79954581 1.74 0.95907049 2.64 0.9958547 3.54 0.99979994 7.40 6.80567E-14
0.85 0.80233746 1.75 0.95994084 2.65 0.99597541 3.55 0.99980738 7.50 3.18634E-14
0.86 0.80510548 1.76 0.9607961 2.66 0.99609297 3.56 0.99981457 7.60 1.47660E-14
0.87 0.8078498 1.77 0.96163643 2.67 0.99620744 3.57 0.99982151 7.70 6.77236E-15
0.88 0.81057035 1.78 0.96246202 2.68 0.99631889 3.58 0.9998282 7.80 3.10862E-15
0.89 0.81326706 1.79 0.96327304 2.69 0.9964274 3.59 0.99983466 7.90 0.00000E+00
* Note: For x>=4.0, 1-PHI(x) is given instead.
 Seoul National University Instructor: Junho Song
Dept. of Civil and Environmental Engineering [email protected]

Probability Distribution Models in R base package

Full Name Short Parameters Probability Density/Mass Function Mean Variance

0  p 1  n x
Binomial binom   p (1 − p) ( n− x ) , x = 0,1,, n np np(1 − p)
n integer  x
Geometric geom 0  p 1 p(1 − p) x , x = 0.1,2,  (1 − p) / p (1 − p ) / p 2

0 k  m+n −1
 m  n  m + n  mk m n m+n−k
Hypergeometric hyper m, n, k     , max(0, k − n)  x  min(k , m) m+n
k
m + n m + n m + n −1
 x  k − x  k 
integers
Negative 0  p 1  r + x − 1 r
nbinom   p (1 − p) x , x = 0,1,... r (1 − p) / p r (1 − p ) / p 2
Binomial r integer  x 
x −
Poisson pois 0 e , x = 0,1,...  
x!

Beta beta 0  q, r B(q, r ) −1 x q −1 (1 − x) r −1 , 0  x 1 q / (q + r ) qr / (q + r + 1) / (q + r ) 2

Chisquare chisq 0 x (  −2 ) / 2 e − x / 2 2 −  / 2 ( / 2) −1 , 0 x  2

Exponential exp 0  e − x
, 0 x 1/  1/  2
[(v1 + v2 ) / 2](v1 / v2 ) v1 / 2 x v1 / 2−1 2v22 (v1 + v2 − 2)
F f 0  1 ,  2 , 0 x v2 /(v2 − 2)
(v1 / 2)(v2 / 2)[1 + (v1 / v2 ) x](v1 +v2 ) / 2 v1 (v2 − 2) 2 (v2 − 4)
Gamma gamma 0  a, b b − a (a) −1 x a −1e − x / b , 0 x ab ab2
Lognormal lnorm , 0   x −1 −1 (2) −1 / 2 exp[ −(ln x − ) 2 / 2 2 ], 0 x e (  +0.5
2
)
e ( 2  + 2  ) − e ( 2  +
2 2
)

Normal norm , 0   −1
 (2) −1 / 2
exp[ −( x − ) / 2 ]
2 2
 2
Rayleigh xb −2 exp( − x 2 / 2b 2 ), 0 x (4 − )b 2 / 2
rayleigh 0b b /2
(package ‘VGAM’)
T t 0 (v) −1 / 2 ((v + 1) / 2)(v / 2) −1 (1 + x 2 / v) − ( v +1) / 2 0 v /(v − 2)
Uniform unif ab (b − a ) −1 , a xb ( a + b) / 2 (b − a ) 2 / 12
a
a −1 x
ax − 
Weibull weibull 0  a,   
, 0 x (1 + a −1 )  2 [(1 + 2a −1 ) −  2 (1 + a −1 )]
   
e

Use dshortname ( ) to compute the probability density/mass function; pshortname( ) to compute cumulative distribution
function; rshortname( ) to generate random numbers; and qshortname( ) to compute the inverse cumulative probability. Use R
help to learn more about these commands.