JSIMP: BN Model and Tool (Version 2.0) for the SIMP project
Contents
1. A typical example of the BN view is shown in Figure 17 it displays a list of the nodes in the selected model item The BN view has four main components Attribute list Toolbar Equation editor not shown in Figure 17 this provides the mechanism for editing NPTs e Top level BN view not shown in Figure 17 this provides the mechanism to link appropriate phase models 4 5 1 Attribute list Each node is represented by a row followed by a number of attributes displayed in columns topped by the name of the attribute Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 28 of 40 The first column in the list is the name of the node This can be changed by clicking once on the name supplying the required text and then pressing the enter key The node names are displayed in one of three different colours e Green This means that the NPT for the node is editable and that it has been fully and correctly specified e Red This means that the NPT for the node is editable but that it has been incompletely and or incorrectly specified e Grey This means that the NPT for the node is not editable Initially all node names will either be green or grey A node will only become red when a user edits the NPT and leaves something unspecified or incorrectly specified The model can still be used however but it means that the NPT used will be the last correctly specified one See Section 4 5 4 below for more information on edit2. change 28 2 rues ye si ae mode button switchs 2a farayalues a to stats mode 3 2 0 Median values 16 g 4 In the Stats mode the 12 scenarios are displayed SS S N in series From first to last Scenario 1 Scenario 2 Scenario 3 Figure 25 Monitor created from a questionnaire with multiple scenarios In the statistics mode the scenarios are displayed in series However because there is no inherent order in the scenarios in JSIMP this fact and that fact that they are joined up can be ignored 4 7 Application modes editor viewer The JSIMP application has two modes of use an editor mode and a viewer mode As discussed in Section 6 development users normally work in editor mode and then create applications that are used in viewer mode by end users The editor mode contains all the functionality of the tool while the viewer mode has a number of functions disabled All of these disabled functions are the ones that allow editing of the questionnaire and BN view specifically e Inthe BN View the buttons for moving nodes up and down and the button for editing the NPT are disabled In addition only the node name and threshold value columns are shown in the attribute list e In the questionnaire view the buttons for creating new questionnaires for moving questions up and down for making questions visible invisible and for deleting questions are all disabled Version 2_0 of SIMP BN Model and Tool 1_0
3. actual cost at any given point the project has associated with it a real cost or budget that is suggested by the project risk attributes This cost is expressed as a percentage deviation from that of the nominal project Thus a value of 15 here would mean that because of the actual project attributes the cost is likely to be 15 more than expected Schedule differential implied by Project Attributes As well as an initial agreed schedule and an actual schedule at any point the project has associated with it a real schedule that is suggested by the project risk attributes This schedule is expressed as a percentage deviation from that of the nominal project Thus a value of 15 here would mean that because of the actual project attributes the schedule is likely to be 15 more than expected Actual Cost Differential This node represents the percentage differential between the actual project budget currently and the budget that is implied by the project risk attributes represented by the node Cost Implied by Project Attribs If this difference is positive Cost Implied by Project Attribs is greater than Actual Cost which suggests that the current budget is not realistic and does not fully take into account the potential risks associated with the project The states of this node are intervals that express the percentage difference between the two parents Actual Schedule Differential This node represents the percentage differen
4. despite the evidence that these are poor Finally close the scenario and open the scenario labelled simp1 This shows how the tool can be used to directly compare different options for a project because you can view multiple scenarios at the same time There are actually two scenarios here To view the second scenario simply click the tab labelled scenario 2 Now select and display the monitors for performance and Cost differential implied by Project Qualities After running Calculate you should see the screen shown in Figure 6 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 13 of 40 Outputs performance reputation Cost differential implied by Proj Qualities Sched differential implied by Proj Attribs Actual Cost Differential Actual schedule Differential Cost and schedule Actual Cost diff above agreed cost Actual schedule diff above agreed schedule Business environment Budget Constraints cash flow Technical Quality Overall Technical Quality Technical Facilities Design Solution Interfacing Obsolescence Key Subsystem Process Quality Process Efficiency Requirements Resources Performance Overall Resource Performance Staff Organisation Effectiveness Planning Facilities motivation Information Subcontract Management procurement Scenario 1 has mainly high or very high project observations whereas scenario 2 has mainly low and very low Not surprisingly the dif
5. is not active in version 2 0 of JSIMP so the corresponding text box should be ignored and its value left at 1 Active Scenarios Delete Project Scenario 1 Project Scenario 4 testing staff quality indicator medium e g staff turnover No Answer x No Answer xl testing staff quality indicator weak No Answer No Answer x testing staff quality indicator strong e g experience No Answer gt No Answer overall testing amp rework management quality No Answer ha No Answer x rework effort No Answer x No Answer rework personnel quality low No answer x prob of fxing defect No Answer No Answer prob of finding defect No Answer x No Answer x testing personnel quality No Answer x No Answer x testing effort No Answer x No Answer x quality of module spec documentation PRE No Answer x No Answer quality of module spec amp documentation POST No Answer gt No Answer gt rework quality No Answer x No Answer x testing process quality this phase No Answer Be No answer xl Project Scenario 1 observations Project Scenario 2 observations Figure 16 The scenario manager with 2 active scenarios In the example in Figure 16 two columns of answer boxes are present the first represents the first scenario while the second is for the last scenario Thus observations placed into the first column will be applied to the first scenario and observations placed into the second c
6. overrun Actual Agreed Sched Diff Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 9 of 40 3 Example using the model The relationship between the nodes described in Sections 2 4 and 2 5 is fairly complex In order to clarify this relationship it is instructive to consider what kind of cost schedule and performance trade off reasoning the model supports To run the following example open the tool and then open the scenario called simp0 You should then see the questionnaire in the main part of the screen as shown in Figure 2 Outputs scenario 1 performance reputation No Answer Cost differential implied by Proj Qualities Sched differential implied by Proj Attribs Actual Cost Differential Actual schedule Differential Cost and schedule scenario 1 Actual Cost diff above agreed cost oO Actual schedule diff above agreed schedule Business environment No Answer Budget Constraints cash flow No Answer Technical Quality scenario 1 Overall Technical Quality his Aer x Technical Facilities 1 Design Solution 4 Interfacing 4 x Obsolescence No Answer Key Subsystem 1 Process Quality scenario 1 Process Efficiency No Answer Requirements No Answer x Resources Performance scenario 1 Overall Resource Performance No Answer x Staff 4 Organisation Effectiveness 1 Planning 2 Facilities 2 motivation 4 x Information 2 Subcontract Management No Answer procurement No Answer Fi
7. recent propagation about each reportable node in the current model file to a CSV comma separated file In the case of multiple scenarios being active on a model only the statistics from the last scenario propagated will be available An example of the format of the CSV data can be seen in Table 1 Table 1 Example of exported statistics Model Node Mean Variance Someni Median Deviation Testing Phase turnover 25 2 059174 1 434982284 2 500000106 Testing Phase indicator weak 25 2 327544 1 525629042 2 500000124 Testing Phase experience 2 5 1 537266 1 239865324 2 500000176 testing staff quality indicator medium e g staff testing staff quality testing staff quality indicator strong e g Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 19 of 40 Testing Phase Testing Phase Testing Phase Testing Phase Testing Phase Testing Phase Testing Phase overall testing amp rework management quality rework effort rework personnel quality residual defects post defects fixed residual defects pre defects found 2 29 2 5 217 8448 32 26082 408 2136 75 9466 0 877849 1 333333 1 333334 144289 2 5432 162 2132421 27744 11 0 936935975 2 500000151 1 154700547 2 50000004 1 154700613 2 500000205 379 8542435 101 9118027 73 70320253 11 40863931 1460 281254 118 7499264 166 565618 31 34826011 4 3 The explorer view The explorer view is situated on the left hand side
8. seleceted monitors up Monitor Move seleceted monitors down Close selected monitors Select deselect all monitors Figure 23 The monitor panel The monitor panel has two main components e The monitor panel toolbar e The monitors Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 35 of 40 4 6 1 The Monitor panel toolbar The monitor panel toolbar contains functions that operate on selected monitors Monitors are selected by clicking on the monitor name in the top left of each monitor Once selected monitors turn green They can be deselected by clicking on the monitor name a second time The functions are e Move selected monitors up This moves all selected monitors one up in the monitor panel e Move selected monitors down This moves all selected monitors one down in the monitor panel e Close selected monitors This closes all the selected monitors e Select Deselect all monitors If one or more monitors are selected then this function will deselect them all Conversely if no monitors are selected then clicking this button will select them all Only a single monitor can be open for a given node at any point If a new monitor is requested for it then that monitor is selected and moved to the top of the list The monitor panel functions are also applied to monitors that haven been detached See Section 4 6 2 for more details on detaching monitors from the monitor panel 4 6 2 Monitors Monitors s
9. the expanded view one of the available tick boxes can be selected Note that clicking again on the same tick box withdraws the observation By entering soft evidence This is done by typing numeric values into the text boxes to the right of the answer tick boxes Any positive number can be entered When the soft evidence is applied to the underlying BN it is normalised the red bar directly under the answer description represents its normalised value Questions that require numeric answers can only be answered by typing answers into the answer text box 4 4 2 The Questionnaire Toolbar The questionnaire toolbar Figure 14 contains a number of buttons which perform functions on the selected questions and question headings Expand Retract selected questions S Display the monitors for selected questions _ Create a new questionnaire heading Move the selected question heading up one Move the selected question heading down one Make the selected questions visible invisible Delete selected items Figure 14 The questionnaire toolbar Expand retract button This toggles the expansion status of the selected items If questionnaire headings are retracted then it will hide all its connected questions until expanded again Create monitor button This creates a monitor for all selected questions The monitors are placed into the monitor panel see Section 4 5 4 Create questionnaire heading button When this butt
10. these nodes are presented below with the corresponding node descriptions e The model now incorporates a powerful representation of the relationship between cost schedule and performance This new representation which emerged from discussion between Astute and RADAR allows trade offs between the three factors to be fully investigated This document describes the main BN in Section 2 In Section 3 we present an example of using the model in the tool while Section 4 is a user manual for the tool Version 2_0 of SIMP BN Model and Tool 1_0 doc 30 July 2003 Page 4 of 40 2 Overview of main JSIMP BN model The top level BN model is shown in Figure 1 this is actually a simplified view of the model since certain dummy nodes are omitted The rest of this section describes the nodes of the model in appropriate groups MORE INFORMATION WILL APPEAR IN FINAL VERSION lt gt Organisation effectiveness C sm D Subcontract management Facilities 8 Resource performance Design Solution Technical Quality Cey subsystem gt Technical Facilities Cost Differential implied Budget Constrain cash flow Process Efficiency Requirements Business environmen Actual schedule differential above agreed Schedule differential implied Actual cost differential above agreed Actual schedule differential Actual cost differential performance Version 2_0 of SIMP BN Model and Tool 1_
11. 0 doc 12 June 2001 Page 5 of 40 Figure 1 JSIMP Main BN Model 2 1 Attributes associated with resource performance Staff Organisation Effectiveness Planning Facilities Information Procurement These nodes are all attributes that correspond to high level risk areas defined by Astute and currently they all have the associated state values Very Low Low Medium High and Very High In the final version of the tool these nodes will all be broken down into the indicators provided buy the ASTUTE team e Subcontract Management This node represents the overall quality of the management of the subcontracting process It can have an impact on both the Key Subsystem risk area and the Procurement risk area e Motivation This node represents the underlying motivation of staff on the project and it can have an impact on the attributes represented by Staff and Management Motivation can be influenced by the effectiveness of the organisation Organisation Effectiveness e Resource Performance This is a derived node that provides an overall score based on observations of the other factors here This node represents the quality of the contribution that the resources make to the project Resource Performance has an impact on Cost Differential Implied by Project Attributes and Schedule Differential Implied by Project Attributes 2 2 Attributes associated with technical quality e Design Solution e Interfacing e Facilities e Obsolescen
12. 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 33 of 40 All nodes will initially have a default NPT At any stage it is possible to return the node s NPT to this state by clicking on the Reset fields to default values hyperlink Conversely if it is decided that the default NPT expression for a node needs to be changed the expression currently shown in the window can be used as the new default by clicking the Set current field values as new defaults This latter option should never really need to be used It should only be used if there is an exceptional reason why the default formula displayed to a user when they begin tailoring the model is perhaps inappropriate or misleading Once the OK button is clicked the new expression will be applied If the NPT edit is successful the window will disappear and the node name will be coloured green in the BN view State expression editing For some nodes that have a single parent it is possible to specify a different expression for each state of the parent node The state expression editor window is shown in Figure 22 below Expressions for prob of fxing defect x Expressions for prob of fxing defect Select a state of the parent rework State very low quality for which to provide an Parameters for functions that only define probabilities for a single state value expression Note You must provide of the parent must be single numerical values i e integers o
13. RADAR Risk Assessment and Decision Analysis Research Group JSIMP BN Model and Tool Version 2 0 for the SIMP project a Siy The RADAR SIMP Tool Risk Assessment and Decision Analysis Research Group Version 1 0 30 July 2003 Abstract The JSIMP tool provides quantitative risk management information at any stage in a major project The tool targeted at project managers and risk managers enables users to enter as little or as much information as they have about a project that can impact on its eventual success or failure At any stage the tool will provide a prediction of a range of key project parameters such as the extent to which the project will go above below schedule above below agreed cost above below a satisfactory level of final system performance likely impact on company reputation etc Because the tool is based on an underlying Bayesian Net BN model predictions for any project parameter are full probability distributions rather point estimates From the full distribution the tool can also show such statistics as means variance etc Prepared by Norman Fenton and Patrick Cates RADAR Risk Assessment and Decision Analysis Research Computer Science Department Queen Mary University of London London E1 4NS Distribution SIMP project partners Version 2_0 of SIMP BN Model and Tool 1_0 doc 30 July 2003 Page 1 of 40 Document Change History Version Date Status De
14. Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 29 of 40 Create Monitor Move node up in the list Move node down in the list Edit the nodes NPT Figure 18 The BN view Toolbar The functions have the following effects e Create monitor This will create a monitor for each selected node The monitor will be appended to the monitor panel see Sections 4 5 4 and 4 6 2 e Move node up in the list This will move all selected nodes one up the list e Move node down in the list This will move all selected nodes one down the list e Edit the node NPT This will bring up the equation editor a modal window in which you can edit the node s NPT The type of form displayed is dependent on the node See Section 4 5 4 for full details In viewer mode only the Create monitor button is available to the user 4 5 3 The top level BN View The top level BN view shown in Figure 19 is available when the top item in the explorer view is selected It displays a representation for each model item as a blue rounded rectangle Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 30 of 40 Toolbar Testing only 1 Link Phase or Testing only 2 la had project model Testing only 3 Figure 19 The top level BN View JSIMP automatically adds a new rectangle for each imported model It is not possible to add rectangles models manually via the BN view Any of the pre defined phase models b
15. ab buttons Once the scenario has been selected its long and short descriptions can be modified simply by clicking on the relevant description and typing in the modifications Scenarios are deleted by firstly selecting the scenario that you wish to delete see above for how to select a scenario then clicking on the delete button situated on the bottom right of the scenario manager 4 5 The work area BN view The BN view is available when a model item is selected in the explorer view and the BN view tab is selected in the work area The only exception is if the top level object in the explorer view is selected in which case a graphical view encompassing all the phase and project models will be displayed Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 27 of 40 Visible Reportable Input Output testing staff quality indicator medium e g testing staff quality indicator weak testing staff quality indicator strong e g t overall testing amp rework management qua Toolbar SERIES Sie Rots residual defects post defects fixed residual defects pri defects found prob of fxing defect prob of finding defect tion PRI quality of module spec amp documentation P rework quality testing process quality this phase I SSNS sngng I Ooni E E E E L E E E EJ E E EJ E L E E E E E E E E E E E E E E E E E E E E E E N Node Attributes Node List Figure 17 The BN View
16. arger domain dependent models e File gt Save Model This saves all the models and links to a single cmp file If this is the first time that the project has been saved during the application session i e since starting JSIMP then a dialog will appear prompting you to specify a path and file name Any open scenarios are NOT saved with this file they must be saved to a separate scenario file see below e File gt Save Model As See the Save Model menu item Here a dialog will always appear promoting you to specify a path and file name e File gt Remove This menu item is further broken down into a list of all the models in the current project Clicking on one of these will remove that model from the project Hence this function enables an individual model to be removed along with any links that existed between it and other models in the current project e File gt Close Model This closes the all the models in the current project e File gt Open Scenario Since scenarios are model dependent this option which opens a dialog listing all scenario scn files is only available if the associated model file has been opened Obviously there will be no scn files unless you have previously saved some see below e File gt Save Scenario As A dialog appears prompting you to select a path and filename to which the scenario should be saved Many different scenarios can be saved for the same model However if the associa
17. at underlie those risks For example instead of a node representing facilities risk we now have a node representing the quality of facilities ranging from low to high The problem with representing risks directly is that a given risk can at best only take on the value of zero This means that the model would not be able fully to support trade offs For example if facilities are known to be an active risk area with a potential impact on the project budget it might be decided to investigate whether an improvement in process efficiency could serve to counteract this negative impact However if process efficiency is treated as a risk such a trade off is not possible the best that can happen as a result of improving process efficiency is that it has no negative impact Only if process efficiency is represented as an attribute with values ranging from low to high can it be allowed to have a positive impact in the model Because the model should be able to support this kind of trade off the current version contains qualities that can have risk associated with them rather than risks per se e State values have been added to all of the nodes in the network Currently for the sake of simplicity the majority of the nodes have the state values Very Low Low Medium High Version 2_0 of SIMP BN Model and Tool 1_0 doc 30 July 2003 Page 3 of 40 and Very High Cost and schedule nodes however are treated differently The state values associated with
18. ce e Key Subsystem These nodes are all attributes that correspond to high level risk areas defined by Astute and currently they all have the associated state values Very Low Low Medium High and Very High In the final version of the tool these nodes will all be broken down into the indicators provided buy the ASTUTE team Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 6 of 40 Technical Facilities This node represents the quality of specifically those facilities that are used during the build process Overall Technical Quality This is a derived node that provides an overall score based on observations of the other factors here This node represents the quality of the contribution that the technical factors make to the project It has an impact on Cost Differential Implied by Project Attributes and Schedule Differential Implied by Project Attributes 2 3 Attributes associated with process quality Process efficiency Requirements These nodes are all attributes that correspond to high level risk areas defined by Astute and currently they all have the associated state values Very Low Low Medium High and Very High In the final version of the tool these nodes will all be broken down into the indicators provided buy the ASTUTE team 2 4 Attributes associated with cost and schedule Agreed cost differential above agreed cost This node represents the percentage differential between the current actual b
19. ct parameter are full probability distributions rather point estimates From the full distribution the tool can also show such statistics as means variance etc The underlying BN model which was obtained using information provided by BAES risk assessors from the ASTUTE project is hidden from users of the tool Users simply enter project risk information via an intuitive questionnaire interface Information they can enter ranges from information about staff quality through to project facilities Users can also enter information about costs and schedules such as the extent to which the project is allowed to overrun Whenever new information is entered the tool will apply Bayesian propagation to update the probability distributions of all the unknown variables The tool has a range of mechanisms for displaying its results Users can also do powerful what if analysis such as investigating how much better worse certain procedures have to be in order to achieve particular performance or cost targets The JSIMP tool is written in Java and has been built on top of Agena s BN architecture called Minerva The underlying BN model has undergone a number of changes Reference 1 The model is still heavily based on the risk categorisation by Astute Reference 2 but it differs in the following ways e Although the whole model is about risk assessment items that were defined as risks by BAES are actually represented by ordinal scale attributes th
20. doc 12 June 2001 Page 38 of 40 Questions that have been defined as invisible in the questionnaire view see Section 4 4 2 will not be displayed in the questionnaire list Similarly nodes that have been defined as invisible in editor mode will not be displayed in the BN view when the tool is used in viewer mode Once a development user has finished all necessary tailoring he she can generate an application by switching the tool to viewer mode This is done by a very simple edit to the text file named minerva properties that is situated in the root directory of the JSIMP application Specifically change the following line uk co agena minerva editable true to uk co agena minerva editable false Also assuming that the development user has created a tailored project specific model file as described in Section 6 called say myproject cmp then change the following line uk co agena minerva defaultModel to uk co agena minerva defaultModel myproject cmp making sure to include the file in the root directory of the JSIMP application when it is give to end users These changes will not take effect until the next time the application is started To revert back to editor mode simply reset the uk co agena minerva editable property to true 4 8 Web enabled version of JSIMP tool JSIMP can be executed over the web using the Webstart technology Full instructions on how to upload the Webstart version of JSIMP to a web server are provid
21. ed in the Webstart zip file that is supplied along with the standard JSIMP distribution zip file Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 39 of 40 4 9 Application FAQ Q How do I create a monitor for a node A Select the node you are interested in from either the questionnaire or the BN view Then click on the create monitor button in the toolbar to the left of the questionnaire or BN view Q I ve entered some answers How do I see the results A Create a monitor for each node question you are interested in then click on the Calculate menu item The model will propagate and display the results in the monitors Q How do I add more questions to the questionnaire A You cannot When a model is imported there will be a question for every node in the questionnaire and no more can be added They can be made invisible when the model is opened in viewer mode by using the visible function in the questionnaire Q I want to delete a link I have created between two models A See section 4 5 3 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 40 of 40
22. ee Figure 24 display probability information for a selected node They are created by selecting the relevant node in the BN or questionnaire view and clicking on the Create monitor button in the connected toolbar see Section 4 5 A newly created monitor will be added to the monitor panel see Section 4 5 4 Monitors can be decoupled from the monitor panel by clicking on the Decouple couple monitor button on the top right of the monitor A decoupled monitor is removed from the monitor panel and placed into its own frame after which it can be resized and minimised like any other window To re couple it simply click again on the Decouple couple monitor button Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 36 of 40 Mode button Decouple Couple monitor Legend Maginal Plot Figure 24 A Monitor A monitor has two view modes The mode is toggled by clicking on the Mode button in the top right hand corner of the monitor see Figure 24 The modes are e Probability Mode This is the default mode shown in Figure 24 This displays the probability distribution for the node from the last calculation see section 4 2 The states are shown on the x axis and the y axis contains a probability between 0 and 1 The median value of the probability plot is shown in brackets after the name in the legend but only after the first calculation If a node has a threshold value set then this is displayed a
23. elected because in this case the questionnaire is empty Tab Bar Figure 11 The work area tab bar The components of the questionnaire view are shown in Figure 12 In this section we explain the following components e The questions list including how to select edit and answer questions e The questionnaire toolbar e The scenario manager Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 21 of 40 es vE ss TestingPhase No current scenario selected Scenario Short Description M Long Description No Long description available Manager Forecast Horizion Time Unit s Testing Network testing staff quality indicator medium e g staff turnover testing staff quality indicator weak testing staff quality indicator strong e g experience overall testing amp rework management quality rework effort rework personnel quality No Question Text has been defined Joo veyiow so eve SEE Expanded LJ b 0 average Question Lj 0 0 hor Lj oo veryhich prob of fxing defect prob of finding defect testing personnel quality testing effort quality of module spec documentation PRE quality of module spec amp documentation POST rework quality testing process quality this phase Defects Network defects found defects fixed residual defects pre residual defects post Answer Combo Box tale Unexpanded Question KAKARAKA Questionnaire Heading Questionnaire Toolbar Figure 12 The questionna
24. elected in the explorer view e The monitor panel This contains a set of monitors where a monitor is an item that displays the probabilities for a given node in the underlying model after propagation More information can be found in Sections 4 5 4 and 4 6 2 It is important to note that by default the JSIMP tool runs in editor mode see Section 4 7 and the users are assumed to be systems users Hence this user manual is written from the perspective of JSIMP systems users rather than end users The systems users will be using the toolset to generate their own applications by tailoring the models and questionnaires appropriately for a specific project Section 4 7 describes how to create a non editable application for end users once the systems user has completed the tailoring It is expected that systems users would provide a modified user manual for their own end users Such a user manual would be based on the relevant subsets of this manual embellished with information about the tailored models and questionnaires It is therefore important that readers first understand the basics of the JSIMP method and models described earlier in this document before engaging with the tool Finally in Section 4 9 there is a brief FAQ on using the application 4 2 The menu bar functions The menu bar contains functions to load save and import models and scenarios It also contains the calculation menu item which is the only mechanism in JSIMP to propaga
25. ferences in performance and Cost differential implied by Project Qualities are very significant To bring the performance in scenario 2 more in line with that of scenario 1 we can increase the budget and schedule as shown in Figure 7 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 scenario 1 No Answer scenario 1 No Answer No Answer scenario 1 No Answer 4 No Answer No Answer No Answer scenario 1 No Answer 5 scenario 1 No Answer AIR Ea aaa a Figure 6 Multiple scenario example scenario 2 No Answer scenario 2 r No Answer x No Answer scenario 2 rl No Answer xh x No Answer No Answer x No Answer El scenario 2 rl No Answer Ela scenario 2 rl No Answer MAKI vj2 MAKI yl2 ba MAKI zj2 vii a i aa ele ON ad cd a ed 1 0 0 8 0 6 0 4 0 2 0 0 oz yur scenario 1 11 25 scenario 2 26 37 4 NSIS N N O TEE ka ne or s oo cS cn scenario 1 2 52 scenario 2 5 43 JSIMP performance OZ oL oe oz or 0e os or 0S u Wy Page 14 of 40 Cost and schedule scenario 1 Actual Cost diff above agreed cost Actual schedule diff above agreed schedule zol JSIMP performance A lL 1a scenario 1 2 52 0 8 scenario 2 0 50 0 6 0 4 a 6 _ ean 0 0 ere tea Seed a cline O Figure 7 Increasing budget and schedule on scenario 2 Versi
26. gure 2 Screenshot for scenario simp0 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 10 of 40 The scenario you have loaded already has some answers entered You can change these simply by using the combo boxes or by entering values for those that have numeric answers In this loaded scenario we have a classic at risk project in the sense that many of the project attributes that we know about have values very low or low Also note that the actual cost above agreed cost has a value 0 meaning effectively that this is a fixed price contract it is not allowed to go over budget The first simple use of the model and tool is as a scorecard type assessment we can see how the model evaluates the overall technical quality and performance quality To do this select the questions overall performance quality and technical quality and display their monitors by clicking on the monitors icon Then select the Calculate Run Calculation menu item When the calculation meaning full Bayesian propagation of the entire model is completed you should see the monitors as in Figure 3 JSIMP Technical Quality H scenario 1 2 36 0 4 0 2 0 0 mer T 5 0 _ JSIMP Resource Performance MAL 1 0 E scenario 1 2 60 1 0 2 0 3 0 4 0 5 0 Figure 3 Distributions for technical quality and overall resource performance The distributions are centred around
27. ing NPTs The following additional attributes are available for nodes e Visible This determines whether the node is visible in the BN view in viewer mode Regardless of this attribute value the node is always visible in edit mode e Reportable This determines whether the node is output when a report is generated e Input A node defined as an input node is configured to receive as input some information from a previous phase model In the pre defined JSIMP model input and output nodes are currently not used e Output A node defined as an output node is configured to send information as input to another phase model see comments for Input above e Threshold Value A threshold value can be defined for any node The threshold value is normally either a desirable value that the user wishes to see the node reach or some limit above or below which the value should not cross Any set threshold values are plotted on the monitors relevant to the node so that users can see at a glance the relationship between the predicted distribution for the node and the threshold value In viewer mode only the uneditable node name and the threshold value will appear in the attribute list 4 5 2 The BN view toolbar The BN view toolbar Figure 18 provides a number of functions that can be applied on the selected nodes in the BN view Nodes are selected by clicking on their names Multiple nodes can be selected by holding done the control key while selecting
28. io 161 40 0 8 0 6 0 4 0 2 0 0 ieee Figure 5 Updated performance prediction after increasing budget However it is still below average There are two reasons for this both of which the model shows clearly 1 We still have poor technical quality and resource performance and these are not fully compensated by the increased resources 2 If we check the prediction for Schedule differential implied by Project Attributes we find that the schedule differential is predicted to be around 30 meaning that we really need to increase the schedule time by 30 So now enter the value 30 in Actual schedule diff above agreed schedule After running Calculate the prediction for performance is now looking good Thus given the poor quality attributes we can still deliver a good system but only by significant cost and schedule overruns Moreover if you check the monitor for reputation you will see that this is tending toward low because despite the respectable system it is being delivered late and over budget Now replace the values entered for Actual schedule diff above agreed schedule and Actual cost diff above agreed cost with 0 s In other words the project cannot go over budget or schedule Suppose that we have a requirement for the performance to be excellent Enter this as an observation in performance What happens now is that both overall resource performance and technical quality increase significantly
29. ire view 4 4 1 The questions list There is a list of questions that corresponds to each particular node for the selected model item Questions are grouped under questionnaire headings The headings are shown in larger black text while the questions are slightly indented and shown in blue The questionnaire headings act simply as a convenient mechanism for grouping related questions together These headings like the questions themselves are editable and users can define their own headings so that they can group questions in a way that they feel is suitable for a particular model or application Questions and headings are selected by clicking on their short description When selected their background colour will change to light blue Clicking on the short description again will deselect the question or heading Multiple items can be selected if the control key is held down during selection Functions applied when questions and headings are selected will then be applied to ALL selected items Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 22 of 40 By default questions appear in the question list unexpanded showing their short description and their answer box For questions whose answer has an ordinal scale measurement such as very low to very high the answer box is a combination box containing the set of possible answers For questions that require a numeric answer the answer box is a text box which takes only
30. k Des t E oe Delete the Selected scenario Scenario short and long description Selected Sceanrio Expand Retract scenario manager Active Scenario Inactive Scenario Figure 15 The questionnaire scenario manager The scenario manager is available for all questionnaires It can be expanded and retracted via the expand retract button that is situated in its top left corner Along the top is a list of the available scenarios shown as tabs The colours are significant e A blue tab denotes an active scenario e A grey tab denotes an inactive scenario e The beige tab denotes the currently selected scenario By default when a model is opened it is configured with just a single scenario Observations entered into questions will be held in that scenario Scenarios can be added by Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 25 of 40 clicking on the Add new button situated in the top right of the scenario manager When it is clicked the new scenario will appear at the end of the scenario tabs Scenarios are toggled active and inactive by clicking on their connected tab button Active scenarios are then made available in the questionnaire for answering Inactive scenarios are not available in the questionnaire Figure 16 shows a questionnaire and scenario manager configured with 4 scenarios the first and last of which are active Note that the Forecast Horizon functionality
31. low in each case although these is still a small probability in each case that it could be higher or lower However much more interesting than these predictions are the predictions for the output nodes For example select the monitors Cost differential implied by Proj Qualities and performance and again run Calculate You should now see the monitors in Figure 4 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 11 of 40 JSIMP Cost Diff Implied by Proj Qualities 1a scenario 1 23 71 08 0 6 0 4 0 2 0 0 a A NE S oe e ee S N a O a O O O 4 D JSIMP performance 1 0 08 scenario 1 7 50 0 6 0 4 0 2 0 0 Figure 4 Predicted distributions for cost differential implied by project attributes and performance These predictions bring bad news The cost differential says that based on the project properties the cost overrun is likely to be between 20 and 30 to compensate for the poor qualities However we know the actual cost is fixed Hence the performance is likely to be very low Now remove the actual cost constraint and enter the value 30 for Actual Cost diff above agreed cost in other words we increase the budget by 30 over the original agreed budget After running Calculate the performance prediction improves as shown in Figure 5 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 12 of 40 IMP performance scenar
32. numeric key presses The expanded version of a question see Figure 13 displays more details about the question including guidelines on when how and if it should be answered Non numeric questions will also display a list of possible answers The answers consist of a tick box for answering the question with a single answer a text box for soft evidence and a short description of the answer To expand a question select the question and then click on the Expand Retract button on the questionnaire toolbar the button at the top of the toolbar Question in the unexpanded view mode rework personnel quality low Selecting the question s then clicking on the expand retract button on the toolbar will expand and retract the selected question s Short Question LJ Answer Combo Box v rework personnel ae ae low Py No Question T t has been defined gt Allowable Answer Long Question Soft evidence answer text box Hard Evidence tick box Figure 13 The question component expanded and retracted Note that both discrete and numeric questions will display a longer description when expanded Questions can be answered via a number of mechanisms In the case where the question has a non numeric answer e An answer can be selected from the combo box The selection of No answer will withdraw any current observation for that question Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 23 of 40 In
33. of the application window It contains a complete list of all the models that are contained within the top level model at any point entir B odel Top Level Item Selected Item B 2nd Testing Phase B 3rd Testing Phase E 4th Testing Phase Figure 10 The explorer view Clicking on an item in the tree view will select the item Full details of that item will subsequently be displayed in the work area on the right hand side of the application window Either the BN view or the questionnaire view will be shown depending on which of the corresponding tabs is selected in the work area Any item in the explorer view can be renamed this is crucial if you want to create your own tailored version of an existing model To rename an item you have to click on it once and then leave the cursor over the item You will then be able to type in text to change the name Changes are committed only once you have pressed the enter key Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 20 of 40 If the top level item is selected see Figure 10 then a view of the top level model will be displayed in the work area When a pre defined model is loaded for the first time this will appear empty 4 4 The work area questionnaire view If the questionnaire tab is selected in the work area see Figure 11 then the work area will display a questionnaire for the item selected in the explorer view except if the top level item is s
34. olumn will be placed into the last scenario There are no restrictions on how many of the scenarios can be active in the scenario manager at any point although screen real estate limits what is reasonable It is useful to note that the order of the columns in the questionnaire is equivalent to the order of the active scenarios For example if we made scenario 2 active in Figure 16 then there would be three answer columns in the questionnaire such that the middle column would now be connected to scenario 2 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 26 of 40 The scenarios are independent of each other Entering observations into one of them in no way modifies the others Observations supplied for questions via their expanded view are applicable only to the left most active scenario This is highly relevant if you wish to enter soft evidence In this case you must ensure that the scenario into which you wish to enter the evidence is the only active scenario The short and long descriptions of the scenario can be modified To do this the relevant scenario must first be selected Scenarios are selected by moving the mouse over the connected scenario button It is crucial to note that NO mouse click is required for selection mouse clicks as discussed above are for making scenarios active inactive The selected scenario is shown in beige while its details are present in the large beige rectangle directly under the scenario t
35. on 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 scenario 2 Page 15 of 40 4 JSIMP TOOL USER MANUAL 4 1 Introduction This appendix describes all the main components of the JSIMP tool which are shown in the JSIMP application window in Figure 8 and how to use them Menu Bar Work Area Explorer View Monitor Panel Of aterase mM crammed Mae CIEMEN CITE NICHI y ret ven Rey reer eer eee v vere ry ee ee ead heimat 2 ay aeee ome Cone eres A Evan amuwens pay rmb the nawt pra py net ped hoa ber wane vvutps eaten jee Let heel etd nemn ot pdr ynsdsssddimi tH wate let me baak POT inw pa kam nmre a ERCA OI CECI DICA O fena Eat ua eiren rore sann aeania iaasa Figure 8 The JSIMP interface overview The main components are e The menu bar This has File Tools Calculate and Help menus Section 4 2 provides the details e The explorer view This contains a list of models which are instances of the project level net and phase level nets that have been loaded or imported into the overall model More information about this can be found in Section 4 2 e The work area This has two views there are tabs for switching between the two namely a questionnaire view described in Section 4 4 and a BN view described in Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 16 of 40 4 5 In either case the work area contains the associated details of the model currently s
36. on is pressed a new questionnaire heading is added at the end of the questionnaire All selected items will be deselected and the new questionnaire heading selected Move items up button This moves all selected items one up the list Questions can be moved between questionnaire headings using this function Move items down button This moves all selected items one down the list Questions can be moved between questionnaire headings using this function Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 24 of 40 e Toggle visibility This toggles the visibility of the selected components Questions that are defined as being invisible will be displayed with grey short question text rather than blue Invisible items will not be shown at all when the model is loaded in viewer mode see Section 4 7 Normally synthetic nodes are set as invisible by default e Delete selected items In version 2 x this function ONLY deletes empty questionnaire headings In order to do this select the questionnaire heading that you want to delete it must have no questions under it then click on the delete button 4 4 3 The questionnaire scenario manager The scenario manager shown in Figure 15 allows the user to manipulate many scenarios at the same time and to plot the results on the same graph via the monitors for comparison purposes g sE e SS i Add new SS ae Scenario PA Scenar io A lt Enter short defcriptioh here gt
37. op up menu Note that in version 2 x of JSIMP you will not be asked for confirmation when you delete a link or model via this mechanism 4 5 4 The NPT editor The NPT editor allows users tailoring models to modify the NPTs of certain nodes As mentioned above the way in which the NPT for a particular node can be edited and indeed whether it can be edited at all depends on the type of node Prior probability editing For leaf nodes it is possible to specify the prior probabilities directly When a leaf node is selected and the NPT editor button clicked from the BN view the NPT editor window appears as shown in Figure 88 below CNPT for rework effort X NPT for rework effort very low 0 111111 low 0 222222 State Names average 0 333333 tabi high 0 222222 very high 0 111111 3 Cancel vf OK Figure 20 The prior probability editor The probabilities can be modified as required by changing the values in the text boxes Note that you do not have to ensure that the probabilities sum to one when you apply the NPT Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 32 of 40 changes by clicking the OK button the probabilities will be normalised automatically The resulting NPT can be checked by opening the NPT editor again for the selected node Node expression editing For some nodes it is possible to provide an expression that defines the entire NPT The NPT editor window for nodes of this kind looks like that
38. r decimals expressions for all states State Equation low Mean State List average Variance high very high Reset fields to default values Set current field values as new defaults x Cancel Reset functions Figure 22 The state expression editor The principle is similar to the node expression editor described earlier When a parent state is selected on the left of the screen an expression editor corresponding to that parent state value is shown on the right Expressions must be supplied for all parent state values before the NPT can actually be updated If all expressions are complete and correct the state labels on the left should all be green As with the node expression editor the window contains hyperlinks that allow the default NPT values for the node to be restored and set Note however that clicking on these links will affect the entire NPT not just the expression for the currently selected parent state Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 34 of 40 4 6 The Monitor Panel The monitor panel Figure 23 is situated on the right hand side of the application window Whenever monitors are created via the BN and Questionnaire view they are placed onto the monitor panel The monitor panel takes up quite a lot of screen real estate For this reason it can be minimised and maximised via the arrow button situated at the top left of the panel Minimise Maximise button Toolbar Move
39. s a vertical line at the relevant point on the x axis e Statistics Mode This displays a range of summary statistics for the node see Figure 25 including the mean median the lower 0 25 percentile and the upper 0 75 percentile The 0 5 confidence interval is shaded blue between the upper and lower bound if there is more than one plot on the graph see below If a monitor is created where the scenario manager see Section 4 4 3 has more than one scenario then the resultant monitor will have a set of data for each scenario Figure 25 displays an example of this Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 37 of 40 PA 2 Monitor created with a distribution for each scenario configured in in the questionnaire gak 1 Select the node and ng e g experience click on the create ent quality monitor button high El Scenario 1 1 63 residual defects post parane aga defects fixed residual defects pre 0 4 i defects found 02 E prob of fxing defect No Answer bat 0 0 _ E PREC 08 08 _ verylow low average high veryhigh testing personnel quality No Answer mn testing effort high E quality of module spec documentation PRE very high N z No quality of module spec amp documentation POST high eno answer rework quality N ainmin _ N i testing process quality this phase f sting_only testing effort lies os Bi Lower Perc spres 3 Clicking the
40. scription 0 1 29 07 2003 Review Initial version for review by RADAR group 1 0 30 07 2003 Final Minor edits before release References 1 Astute Risk Management BBN Version 1 4 RADAR Group Queen Mary University of London June 2001 2 Extended Astute Risk Information Astute 2001 3 Building large scale belief networks Neil M and Fenton N Knowledge Engineering Review 15 3 257 284 2001 Contents Mammo F043 079 b c10 1 8 Deresepe Ret army a Rene re ge rec Mtr ie eee meio st ne RnR eee ee eee 3 2 Overview of main JSIMP BN model siveccite a eared ee eee 5 2 1 Attributes associated with resource performance c cccceesseceseceeeeeeeeeeseeesecneeeeees 6 2 2 Attributes associated with technical quality 00 0 0 cecceccceesessseceteceeeeeeeeesseeesseenteenees 6 2 3 Attributes associated with process quality cccccccesccescceesceesseceeceseeeeeeeeseecsaeenteenees T 2 4 Attributes associated with cost and schedule 0 ccc ce cceccccescecesececeseceesseceesseeeesseeees 7 2 5 Outputs predicted attributes sssseeseeeeesseeseesseesseseessessesstsseessesstsstesseeseesresseesreseessee 8 3 Example using the WMO ela 4s ssnee vend sa hedacec abcess sanaver vancencis vasues cvs eed eas atedengs oats 10 4 JSIMP TOOL USER MANUAL rrenan aaran estara eE EA A A EA aa i aA 16 4 1 Tntrod ctioin sea aa a E A el eee yh a 16 4 2 The menu bar functions ssenssenesnneosseesseesseossnsesssessstessen
41. shown in Figure 21 below Expression for overall testing amp rework management quality Expression for overall testing amp rework management quality Function parameters take the form of standard mathematical expressions and can include node names available by right clicking in the parameter s text field If a parameter is badly formed the text field will have a red border You can find out the problem by holding the mouse over the field Function Type foma oo O Mean Variance Reset fields to default values Set current field values as new defaults 3 Cancel wv OK Figure 21 The node expression editor The function type should first be selected the parameters that are appropriate for the selected function type can then be entered The parameter values specified can themselves be compound mathematical expressions and they can also refer to the node s parent s if required When entering a parameter expression at any stage the name of a parent can be included by right clicking on the text box and choosing the appropriate item from the menu that pops up To specify a non statistical mathematical expression the function type needs to be Arithmetic When this type is selected only one parameter text box will be shown If there are errors in a parameter expression the text box will be surrounded by a red border More information about an error can be displayed by moving the mouse over the text box Version
42. ssessssressseesseessensssessereesnes 17 4 3 Theexplorer sie Wen rinn sort tunn an e E taal E ate aa anand 20 4 4 The work area questionnaire View iici cccs dis serccssiccns casa ckevacte cts Mis epesticunscassegersaces 21 4 5 The work at a BN View sesissscscciay cdedaasasstadadaasatiasavea EATE AECT AETA AES aan 27 4 6 The Monitor Panela denaii E A Waa neat uae ee 35 4 7 Application modes editor Viewe ccccesccesscessseeseeceseceseceeeeeescecaeceeeseneenaeeenaeees 38 4 8 Web enabled version of JSIMP 100 leis tnsacidiness save wecatdes mane nanchtantschdeiciosedds eakaeetene 39 49 Application FAQ Seere eae E a A atrceids Geen casas EEE 40 Version 2_0 of SIMP BN Model and Tool 1_0 doc 30 July 2003 Page 2 of 40 1 Introduction The JSIMP tool produced by Queen Mary as part of the SIMP project provides quantitative risk management information at any stage in a major project The tool targeted at project managers and risk managers enables users to enter as little or as much information they have about a project that can impact on its eventual success or failure At any stage the tool will provide a prediction of a range of key project parameters such as the extent to which the project will go above below schedule above below agreed cost above below a satisfactory level of final system performance likely impact on company reputation etc Because the tool is based on an underlying Bayesian net BN model predictions for any proje
43. te the underlying models The menus can be seen in Figure 9 File Tools Calculate Help Open Model File Import Model File Save Model File Save Model As File Remove gt Close Model Open Scenario File Tooke Calculate 7 Run Calculation BN Images gt Clear entered data gt Generate Report Save Scenario As File Export statistics Close Scenario Exit Figure 9 The menus Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 17 of 40 4 2 1 The File Menu Functions The file menu functions are concerned with model files and scenario files e File gt Open Model A model file should be viewed as a project it is a collection of one or more BN models which may or may not be linked This menu item opens a dialog listing available model files which have the extension cmp When using the Open Model operation any existing models will be closed This is in direct contrast to the Import Model function see below which simply adds the selected model to the collection of existing models e File gt Import Model This option which requires users to have opened a model file opens a dialog listing available model files When a file is selected it is imported into the current model and all its component BN models are added to those in the current model Importing does NOT close any of the existing models Hence this is the main mechanism for building up l
44. ted model is changed then any scenarios previously saved for it may be invalid Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 18 of 40 4 2 2 File gt Close Scenario This will remove all observations from all scenarios currently defined in the models File gt Exit Closes down the application NOTE You will NOT be prompted as to whether you wish to save or not The Calculate Menu Functions There is just one crucial function available This function propagates all the open models All open monitors will update during the propagation Monitors that are created after the propagation will NOT show the correct probabilities until another calculate has been performed 4 2 3 The Tools Menu Functions Tools gt BN Images This enables the users to view as a graph any of the pre defined BNs i e the project level model and the various phase models The images are picked up directly from the root directory of the application Tools gt Clear Entered Data Removes all the observations in scenarios connected to the specified model Tools gt Generate Report Creates an HTML report detailing the models and their nodes The long and short description of each node is reported along with any answers defined for it in the connected scenarios Nodes can be made reportable or non reportable from the BN view see Section 4 5 Tools gt Export Statistics This option sends current statistical information from the most
45. tial between the actual project schedule currently and the schedule that is implied by the project risk attributes represented by the node Schedule Implied by Project Attributes If this difference is positive Schedule Implied by Project Attributes is greater than Actual schedule which suggests that the current schedule is not realistic and does not fully take into account the potential risks associated with the project Performance This node represents the overall performance of the system being built by the programme Performance is directly influenced by two variables the difference between the actual project schedule currently in place and the schedule that is implied by the project risk attributes Actual Implied Sched Diff and the difference between the actual project budget currently in place and the budget that is implied by the project risk attributes Actual Implied Cost Diff None of the project qualities e g resources requirements Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 8 of 40 process efficiency affect performance directly their effects are mediated by Actual Implied Sched Diff and Actual Implied Cost Diff e Reputation This node represents the reputation of BAE Systems in the commercial environment BAE Systems reputation can be influenced both by the performance of the system built on the Astute programme Performance and by the perceived budget overspend Actual Agreed Cost Diff or schedule
46. udget and the budget agreed at the start of the project If this difference is positive say 15 then Agreed cost is higher than Actual cost by 15 which implies that the project is running within the planned budget Conversely if the difference is negative then Agreed cost is lower than Actual cost which suggests that the project is not running within the planned budget This approach allows the BN to be used for projects of all sizes and costs but without the need for detailing a specific range of potential budgets Agreed schedule differential above agreed schedule This node represents the percentage differential between the current actual schedule and the schedule agreed at the start of the project If this difference is positive say 15 then Agreed schedule is higher than Actual schedule by 15 which implies that the project is running within the planned timeframe Conversely if the difference is negative then Agreed schedule is lower than Actual schedule which suggests that the project is not running within the planned timeframe This approach allows the BN to be used for projects of all sizes and costs but without the need for detailing a specific range of potential timeframes Budget constraints cash flow Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 7 of 40 Business environment 2 5 Outputs predicted attributes Cost differential implied by Project Attributes As well as an initial agreed cost and an
47. ut NOT the project level model may be linked In general model A may be linked to model B if A has an output node of the same type as an input node in B each phase net has two pre defined input nodes and two pre defined output nodes Links are created via the link tool available from the tool bar When the link tool is clicked a new link will appear at your mouse cursor You should then click somewhere in an empty part of the canvas A new selected link will appear with two control points light grey boxes at either end of the arrow Clicking and holding down the mouse pointer inside a control point allows you to drag it around the canvas Releasing the mouse pointer leaves that end of the arrow at that location If an end is dragged and released inside a model the blue rectangles then that end will become anchored to that object Links are created by dragging the two ends of an arrow into different models The source output model must be connected to the line end of the arrow while the target the one that receives the output of the source should be connected to the arrow end Links are created in JSIMP by matching the output nodes of the source model to the input nodes of the target These attributes are defined in the BN view See Section 4 5 Version 2_0 of SIMP BN Model and Tool 1_0 doc 12 June 2001 Page 31 of 40 Links and models can be removed by right clicking on the selected item then selecting delete from the subsequent p
Download Pdf Manuals
Related Search