Lxerc|se 4

A new pro[ecL has !"#$!%# novelLy for Lhe soware suppller LhaL ls golng
Lo execuLe lL and Lhus glven a &'()&!* raung on Lhls accounL for
precedenLedness (8LC). uevelopmenL exlblllLy (lLLx) ls #+,$! -)%-, buL
requlremenLs may change radlcally and so Lhe rlsk resoluuon exponenL
(8LSL) ls raLed *'.. 1he developmenL Leam ls all locaLed ln Lhe same omce
and Lhls leads Lo Leam coheslon (1LAM) belng raLed as "#$/ -)%-. 8uL Lhe
soware house as a whole Lends Lo be very lnformal ln lLs sLandards and
Lhe procedures and Lhe process maLurlLy drlver (MA1) has Lherefore
been glven a raung of "#$/ *'..
1ASk:
CalculaLe Lhe !"#$% '#"()* (!') ln Lhls case
LsumaLe Lhe overall %+)*( lf Lhe slze of Lhe appllcauon ls esumaLed as 3300
llnes of code
156
Send your solution by email [email protected]
with a subject [Proj. Mng.] Exercise 4
Bring a paper copy and give it to the
lecturer before Lecture 5 (17. Mar.)
OR
CCCCMC II
Scale facLor values
recedentedness (kLC)
uegree of precedence or slmllar
pasL cases
Deve|opment ex|b|||ty (ILLk)
1he number of dlerenL ways Lo
meeL requlremenLs
Arch|tecture]r|sk reso|unon (kLSL)
uegree of uncerLalnLy abouL Lhe
requlremenLs
1eam cohes|on (1LAM)
ls Lhe Leam dlspersed
rocess matur|ty (MA1)
SLrucLure and organlsauon of Lhe
developmenL process
157
Dr|ver Very |ow Low Nom|na| n|gh Very h|gh Lxtra h|gh
kLC 6.20 4.96 3.72 2.48 1.24 0
ILLk 3.07 4.03 3.04 2.03 1.01 0
kLSL 7.07 S.6S 4.24 2.83 1.41 0
1LAM 3.48 4.38 3.29 2.19 1.10 0
MA1 7.80 6.24 4.68 3.12 1.36 0
So|unon to Lxerc|se 4
Scale facLor and LorL
!"#$% '#"()* = , - 0.01 x z (%./)0%0( 1*23%* *#405!) =
= 0.91 + 0.01 x (3.72 + 0 + 3.63 + 1.10 + 7.8) = 1.0927
%+)*( = A (!26%)
(!')
=
= 2.94 x 3.3
1.0927
= 11.3371 (person-monLhs)
158
Dr|ver kanng Va|ue
kLC nomlnal 3.72
ILLk exLra hlgh 0
kLSL low 3.63
1LAM very hlgh 1.1
MA1 low 7.8
159
M1A1.03.174 ro[ect Management
LecLure 3: 8lsk ManagemenL
nughes 8., Couere|| M., 1'2.!$# 3$'4#5, 6!&!%#(#&,, 7)2-
89):'&, McCraw-Plll Plgher educauon, 2009
ChapLer 7
160
8lsk
An uncerLaln evenL or condluon LhaL, lf lL occurs, has a posluve or
negauve eecL on a pro[ecL's ob[ecuves
What are pro[ect ob[ecnves?
1he change of exposure Lo Lhe adverse consequences of fuLure
evenLs
lL relaLes Lo future
Whlch ls uncerLaln
lL lnvolves cause and eect
161
Causes Lects
SLa lnexperlence
Lack of Lop managemenL commlLmenL
new 1echnology
users uncerLaln of Lhelr requlremenLs
1esung Lakes longer Lhan planned
lanned eorL and ume for acuvlues exceeded
ro[ecL scope lncreases
1lme delays ln gemng changes Lo plans agreed
162
0. Select project
5. Estimate effort for
each activity
3. Analyse project
characteristics
4. Identify the
products and activities
6. Identify activity risks
7. Allocate resources
8. Review/publicise
plan
1. Identify project
scope and objectives
2. Identify project
infrastructure
9. Execute plan
10. Lower level
planning
Lower-level
detail
Review
For each
activity
CaLegorles of rlsk
163
Actors
Technology Structure
Tasks
R1. High staff turnover leads to expertise of
value to the project being lost
R3.
Appropriateness
of the technology
and to possible
faults within
them, especially if
they are novel
R2. The
implementation
might need user
participation in
some tasks, but
the responsibility
for managing the
users contribution
might not be clearly
allocated
R4. The complexity of the work might lead
to delays because of the additional time
required integrate the large number of
components
R5. If a development technology is novel then the developers might not
be experienced in its use and delay results
A framework Lo deal wlLh rlsk
8lsk ldenucauon
8lsk analysls and prlorlusauon
8lsk plannlng
8lsk monlLorlng
164
8lsk ldenucauon
Check ||st
165
ersonne| Shorua||s
use Lop LalenL
Leam bulldlng
Lralnlng
Unrea||snc schedu|es]budgets
mulusource esumauon
deslgnlng Lo cosL
requlremenLs scrubblng
Deve|op|ng the wrong Sohware
funcnons
beuer requlremenLs analysls
organlzauonal/operauonal analysls
Deve|op|ng the wrong User
Interface
proLoLypes, scenarlos, Lask analysls
Go|d |anng
requlremenLs scrubblng
cosL beneL analysls
deslgnlng Lo cosL
Connnu|ng stream of reqts
changes
hlgh change Lhreshold
lnformauon hldlng
lncremenLal developmenL
Shorua||s |n externa||y furn|shed
components
early benchmarklng
lnspecuons, compaublllLy analysls
Shorua||s |n externa||y performed
tasks
pre-award audlLs
compeuuve deslgns
kea|-nme performance shorua||s
LargeLed analysls
slmulauons, benchmarks, models
Stra|n|ng computer sc|ence
capab|||nes
Lechnlcal analysls
checklng sclenuc llLeraLure
8lsk assessmenL
LlsL of rlsks ls poLenually endless
ulsungulsh Lhe damaglng and llkely rlsks
*2!7 %./)!8*% = (/)(%04#$ 1#9#5%) x (/*):#:2$2(; )' )""8**%0"%)
A pro[ecL depends on a daLa cenLre vulnerable Lo re
lf a re occurs a new compuLer congurauon could be
esLabllshed for S00.000 euro
Where Lhe compuLer ls locaLed Lhere ls a 1 ln 1000 chances of
a re acLually happenlng.
robablllLy of 0,001
1he rlsk exposure ls 300.000 x 0,001 = 300 euro
lf 1000 companles, all ln Lhe same posluon, each conLrlbuLed 300 euro,
Lhen Lhere would be enough money Lo cover Lhe cosL of recovery (1 ln
1000 chance of re occurrence)
166
8lsk assessmenL
LvaluaLe rlsk llkellhood and rlsk lmpacL on Lhe relauve scale ln Lhe
range 0 Lo 10
llgures are sub[ecuve - dlerenL analysLs mlghL plck dlerenL
numbers
167
kef nazard L|ke-
||hood
Impact k|sk
k1 Changes Lo requlremenLs speclcauon durlng codlng 8 8 64
k2 Speclcauon Lakes longer Lhan expecLed 3 7 21
k3 SlgnlcanL sLa slckness aecung crlucal paLh acuvlues 3 7 33
k4 SlgnlcanL sLa slckness aecung non-crlucal paLh acuvlues 10 3 30
kS Module codlng Lakes longer Lhan expecLed 4 3 20
k6 Module Lesung demonsLraLes errors or declencles ln deslgn 4 8 32
8lsk assessmenL
robab|-
||ty |eve|
kange
Plgh CreaLer Lhan 30 chances of
happenlng
SlgnlcanL 30-30 chances of happenlng
ModeraLe 10-29 chances of happenlng
Low Less Lhan 10 chances of
happenlng
168
Impact
|eve|
kange
Plgh More Lhan 30 above budgeLed
expendlLure
SlgnlcanL 20 Lo 29 above budgeLed
expendlLure
ModeraLe 10 Lo 19 above budgeLed
expendlLure
Low WlLhln 10 of budgeLed expendlLure
Tolerance line
8lsk plannlng
k|sk acceptance
uo noLhlng
uamage ls less Lhan cosL
k|sk reducnon and
m|nganon
1ake precauuons LhaL reduce
Lhe probablllLy of Lhe rlsk
k|sk reducnon - auempLs Lo
reduce Lhe llkellhood of Lhe rlsk
occurrlng
k|sk m|nganon - acuon Laken Lo
ensure LhaL Lhe lmpacL of Lhe
rlsk ls lessened when lL occurs
k|sk avo|dance
Some acuvlues may be
prone Lo accldenL LhaL ls
besL Lo avold Lhem
alLogeLher
lf afrald by sharks - do
noL go lnLo waLer
k|sk transfer
8lsk ls Lransferred Lo
anoLher person or
organlsauon
CuLsource Lo an ouLslde
agency
169
8lsk managemenL
Dec|d|ng on r|sk acnons
lor each rlsk speclc acuons needs Lo be planned
CounLermeasures musL be cosL eecuve
8lsk reducuon leverage:
kkL= (kL
before
- kL
aher
)](cost of r|sk reducnon)
Above 1 - Lhe reducuon ln rlsk exposure achleved by a measure
ls greaLer Lhan lLs cosL
200.000 euro Lo replace a hardware congurauon , 1 chance of re.
8lsk exposure would be 2.000 euro
lnsLalllng re alarm aL a cosL of 300 euro would reduce Lhe change of re
Lo 0,3. new rlsk exposure would be 1.000 euro
88L = 2. 1he acuon would Lherefore be deemed worLhwhlle.
170
Creaung and malnLalnlng Lhe rlsk reglsLer
171
Lxerc|se S
Appllcauon of L81
172
11: CalculaLe Lhe expecLed durauon and sLandard devlauon for each acuvlLy
12: ldenufy Lhe crlucal paLh
13: CalculaLe Lhe sLandard devlauon for each pro[ecL evenL (node)
14: CalculaLe 6 values for each node LhaL has a LargeL daLe
1S: ConverL 6 values Lo probablllues
WhaL ls a probablllLy Lo compleLe acuvlLy C by week 30?
WhaL ls a probablllLy Lo compleLe Lhe pro[ecL wlLhln 40 weeks?
Acnv|ty Depends
on
Cpnm|snc
nme (!)
Most ||ke|y
nme (()
ess|m|snc
nme (;)
Lxpected
nme (()
Standard
dev|anon (<)
A - 8 (weeks) 10 (weeks) 12 (weeks)
8 A 10 13 20
C 8 3 7 9
D - 8 10 12
L u, C 3 6 9
Send your solution by email [email protected]
with a subject [Proj. Mng.] Exercise 5
Bring a paper copy and give it to the
lecturer before Lecture 6 (24. Mar.)
OR
L81
rogram Lva|uanon and kev|ew 1echn|que
Assess uncerLalnLy surroundlng esumaLes of Lask
durauon
8equlres:
Most ||ke|y nme
1he ume we would expecL Lhe Lask Lo Lake under normal
clrcumsLances, 9
Cpnm|snc nme
ShorLesL ume ln whlch we could expecL Lo compleLe Lhe acuvlLy, #
ess|m|snc nme
1he worsL posslble ume, allowlng for all reasonable evenLuallues, :
A slngle expected duranon
(
e
= (#+49+:) / 6
173
Lxample
174
Acnv|ty Cpnm|snc
(!)
Acnv|ty duranon (weeks)
Most ||ke|y (()
ess|m|snc
(;)
Lxpected
duranon (,)
A 3 6 8
8 3 4 3
C 2 3 3
D 3.3 4 3
L 1 3 4
I 8 10 13
G 2 3 4
n 2 2 2.3
Lxample
175
Acnv|ty Cpnm|snc
(!)
Acnv|ty duranon (weeks)
Most ||ke|y (()
ess|m|snc
(;)
Lxpected
duranon (,)
A 3 6 8 6,17
8 3 4 3 4
C 2 3 3 2,83
D 3.3 4 3 4,08
L 1 3 4 2,83
I 8 10 13 10,3
G 2 3 4 3
n 2 2 2.3 2,08
L81
Us|ng expected duranons
uene forward pass Lhrough acuvlLy
neLwork
1he calculaLed evenL daLes are Lhe daLes whlch we expecL
Lo achleve Lhose evenLs
lL Lakes an uncerLalnLy
of Lhe real world
lL also focuses on
uncerLalnLy of
acuvlLy durauon
176
L81
Acnv|ty standard dev|anon
A quanuLauve measure of Lhe degree of uncerLalnly of an acuvlLy
durauon esumaLe may be obLalned by calculaung Lhe sLandard
devlauon
! = (: - #) / 6
177
Acn-
v|ty
Cpnm|snc
(!)
Acnv|ty duranon (weeks)
Most ||ke|y (()
ess|m|snc
(;)
Lxpected
duranon (,)
Standard dev|anon
(<)
A 3 6 8 6,17
8 3 4 3 4
C 2 3 3 2,83
D 3.3 4 3 4,08
L 1 3 4 2,83
I 8 10 13 10,3
G 2 3 4 3
n 2 2 2.3 2,08
L81
Acnv|ty standard dev|anon
A quanuLauve measure of Lhe degree of uncerLalnly of an acuvlLy
durauon esumaLe may be obLalned by calculaung Lhe sLandard
devlauon
! = (: - #) / 6
178
Acn-
v|ty
Cpnm|snc
(!)
Acnv|ty duranon (weeks)
Most ||ke|y (()
ess|m|snc
(;)
Lxpected
duranon (,)
Standard dev|anon
(<)
A 3 6 8 6,17 0.3
8 3 4 3 4 0.33
C 2 3 3 2,83 0.17
D 3.3 4 3 4,08 0.23
L 1 3 4 2,83 0.3
I 8 10 13 10,3 1.17
G 2 3 4 3 0.33
n 2 2 2.3 2,08 0.08
L81
1he ||ke||hood of meenng targets
L81 provldes a meLhod Lo esumaLe Lhe
probablllLy of meeung or mlsslng LargeL daLes
CalculaLe Lhe sLandard devlauon of each pro[ecL evenL
CalculaLe Lhe 6 value for each evenL LhaL has a LargeL
daLe
CoverL 6 values Lo a probablllues
179
L81
Lxamp|e assumpnons:
We musL compleLe Lhe pro[ecL wlLhln 13 weeks
AcuvlLy C musL be compleLed by week 10
LvenL 3 represenLs Lhe dellvery of lnLermedlaLe producL Lo Lhe cusLomer
whlch musL Lake place by week 10
180
L81
Ca|cu|anng the standard dev|anon of each pro[ect event
Carry ouL a forward pass uslng Lhe acuvlLy sLandard devlauons (slmllar Lo
expecLed durauons)
1o add Lwo sLandard devlauons - add Lwo squares and Lhen nd Lhe square
rooL of Lhe sum
Lxamp|e:
Lvent 3 depends solely
on acuvlLy 8, lLs
sLandard
devlauon 0,33
Lvent S has Lwo
posslble paLhs
8+L (v(0.33
2
+0.3
2
) = 0.6
and l for whlch 1.17,
sLandard devlauon ls
greaLer Lhan 1.17
181
Ca|cu|anng the = va|ues
lor each node LhaL has a LargeL daLe
LqulvalenL Lo a number of sLandard devlauons (!) beLween
Lhe node's expecLed daLe ((
%
) and LargeL daLe (<)
6 = ( < - (
%
) / !
Lvent 4:
6 = (10-9)/0.33 =
= 1.18867
Lvent S:
6 = (10-10.3)/1.17=
= -0.42064
Lvent 6:
6 = (13-13.3)/1.22=
= 1.2293
L81
182
L81
Convernng = va|ues to probab|||nes
Lvent 4: 6 = 1.8867
= = 3
Lvent S: 6 = -0.42064
= = 67
Lvent 6: 6 = 1.2293
= = 11
183
L81
locuses on Lhe uncerLalnLy of forecasung
SLandard devlauon could be used Lo rank acuvlues
accordlng Lo Lhelr degree of rlsk
We can seL LargeLs
We can esumaLe probablllLy of meeung Lhese LargeLs
8y semng LargeL daLes along Lhe crlucal acuvlLy paLh, we
can focus on Lhose acuvlues poslng Lhe greaLesL rlsk Lo Lhe
pro[ecL schedule
184
Lxerc|se S
Appllcauon of L81
185
11: CalculaLe Lhe expecLed durauon and sLandard devlauon for each acuvlLy
12: ldenufy Lhe crlucal paLh
Acnv|ty Cpnm|snc
nme (!)
Most ||ke|y nme
(()
ess|m|snc
nme (;)
Lxpected
nme (()
Standard
dev|anon (<)
A 8 (weeks) 10 (weeks) 12 (weeks)
8 10 13 20
C 3 7 9
D 8 10 12
L 3 6 9
Lxerc|se S
Appllcauon of L81
Acnv|ty C musL be compleLed
by week 30
ro[ect musL compleLed
wlLhln 40 weeks
186
Acnv|ty Cpnm|snc
nme (!)
Most ||ke|y nme
(()
ess|m|snc
nme (;)
Lxpected
nme (()
Standard
dev|anon (<)
A 8 (weeks) 10 (weeks) 12 (weeks)
8 10 13 20
C 3 7 9
D 8 10 12
L 3 6 9
13: CalculaLe Lhe sLandard devlauon for each
pro[ecL evenL (node)
14: CalculaLe 6 values for each node LhaL has a
LargeL daLe
Lxerc|se S
Appllcauon of L81
187
1S: ConverL 6 values Lo probablllues
WhaL ls a probablllLy Lo compleLe acuvlLy C by week 30?
WhaL ls a probablllLy Lo compleLe Lhe pro[ecL wlLhln 40 weeks?
Acnv|ty Cpnm|snc
nme (!)
Most ||ke|y nme
(()
ess|m|snc
nme (;)
Lxpected
nme (()
Standard
dev|anon (<)
A 8 (weeks) 10 (weeks) 12 (weeks)
8 10 13 20
C 3 7 9
D 8 10 12
L 3 6 9
Send your solution by email [email protected]
with a subject [Proj. Mng.] Exercise 5
Bring a paper copy and give it to the
lecturer before Lecture 6 (24. Mar.)
OR
WhaL dld we learn Loday?
Pow Lo ldenufy and manage rlsks LhaL mlghL
aecL Lhe success of a pro[ecL
8lsk assessmenL and prlorlusauon
uraw up plan s for addresslng Lhese rlsks before
Lhey become problems
LsumaLe rlsk on Lhe acuvlLy neLwork and schedule
188
189