SIX SIG NTGE,Inc for Software evelopment nalytical Hierarchy Process (HP) Simple Excel HP Tool Expert Choice T Tool Overview 2003 Six Sigma dvantage, Inc ll rights reserved FSS: easure 44 FSS: easure 44
nalytical Hierarchy Process (HP) The problem: HP was initially developed by Thomas Saaty at the Wharton School Prioritization, judgments and decisions bog down (a lot) as complexity and number of choices goes up The HP solution: Basic Prioritization 1. Create an efficient set of pair-wise comparisons 2. Synthesize the comparisons to rational priorities 3. (optional) Check the internal consistency in the synthesis 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 8 FSS: easure 8
Basic HP Prioritization How does (the row item) Factors HP Example1.xls Same Factors Simple HP Illustration B C products power prodvec Priorities 1 1 1 1.00 1.00 3.0 3.0 33% B 1.00 1 1 1.00 1.00 3.0 3.0 33% B C 1.00 1.00 1 1.00 1.00 3.0 3.0 33% C Sums 3.0 100% In terms of preference, likelihood, risk, etc Compare with (the column item)? 1 Equal 5 Considerably ore 10 uch ore / No Question 1/5 Considerably Less 1/10 bsolutely Less / No Question #Factors 3 Consistency Index 0.0 Constant for 3 factors 0.58 Inconsistency Ratio 0.0 if Fac2 is stronger, use the reciprocal to indicate the relationship in the top half of the grid One way to represent all the pair-wise comparisons among a set of factors is in a matrix. In this simple example, three factors (, B, C) are assessed. There are 3 comparisons (B, C, BC) that need to be made. That requires only half the matrix, on one side of the diagonal. The other half of the matrix is computed as the reciprocals of the one that directly receives user data. scale using 1 to represent the same rates the comparison of the row item to a corresponding column item at the intersection being evaluated. In this example all comparisons are shown as 1, meaning =B=C. This illustrates that the priorities are all equal and that there is no inconsistency in the set of comparison (i.e. it can be selfconsistent that all three things would be equal). 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 9 FSS: easure 9
Checking for Consistency Fac1 Comparing with Fac2 Simple HP Illustration B C products power prodvec Priorities 1 7 2 14.00 2.41 7.7 3.2 65% B 0.14 1 1 0.14 0.52 1.7 3.2 14% B C 0.50 1.00 1 0.50 0.79 2.5 3.2 21% C Sums 3.2 100% In [importance, preference, likelihood, etc] Fac1 (row) to Fac2 (column) is. 1Equal 5 Considerably ore 10 uch ore / No Question 1/5 Considerably Less 1/10 bsolutely Less / No Question #Factors 3 Consistency Index 0.1 Constant for 3 factors 0.58 Inconsistency Ratio 0.2 if Fac2 is stronger, use the reciprocal to indicate the relationship in the top half of the grid C B = C??? Inconsistency! This case illustrates inconsistency. is evaluated to be a lot more important than B (the 7 in the upper left white cell in the matrix). is assessed to be only a little more important than C (the 2). Note the visual representation of the comparison to the right is a lot higher than B and only a little higher than C. The final evaluation cell indicates that B is equal to C (the 1). This is inconsistent there s no way that all these relationships can coexist. Sometimes HP users show these kinds of inconsistencies in their thinking when they work through a set of comparisons. Shows up in Inconsistency Ratio Rule of thumb: IR > about 0.1 can cause concern 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 10 FSS: easure 10
Simple Excel HP Tool Gathers user input in spreadsheet Computes linear priorities Reports in Pareto Chart Example: Instructions Executive Input: Stratetic Priorities (CTQ's) Prioritization Excel HP Tool.xls Request: The team needs the top strategic business priorities (Project CTQ's) validated and prioritized These weighted and prioritized CTQ's will be the road map for making trade-off decisions in developing the product. How can you provide this input? HP Tool Step 1 Review the list of CTQ's for completeness (if you add any, noted equations need to be updated) Step 2 Fill out the HP by methodically comparing each CTQ to the other ones Step 3 Use the scoring method (more, same, less important) as defined on the HP tab Step 4 Gray cells are calculated and the score is automatically calculated Step 5 Pros ost accurate way Calculated scores Cons ust take a minute to understand HP HP sometimes is laborous 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 45 FSS: easure 45
Excel HP Tool Scale Simple HP 1 - equally important Instructions 5 - ore Important Fill the CTQ's in the rows. They are automatically copied into the column headers 10 - uch more important Fill in bottom half of the matrix and the top calculates itself as the inverse 1/5 - Less Important These criteria and scores are pulled into the ecision atrix weighting column 1/10 - uch less important Note: Grey boxes are calculated fields, do not edit Rank Strategic Business Priorities Is [row] more important than [column]? (>1) Time to market (Technology #1) Time to market (Technology #2) Clinically integrated eets beta customer dates Exceeds the competition Profitable Efficient install Efficient support capability Performance (avg screen to screen & field to field < 1sec) Scalable (to 10,000 users) Customizable 9 Time to market (Technology #1) 10 0.2 1 5 5 1 1 0.2 0.2 5 1 16 Time to market (Technology #2) Fill-in 0.1 0.1 0.1 0.1 5 0.1 0.1 0.1 0.1 1 0.2 4 Clinically integrated 5 10 5 5 5 1 1 1 1 5 1 8 eets beta customer dates 1 10 0.2 5 5 1 1 1 1 5 5 13 Exceeds the competition 0.2 10 0.2 0.2 5 0.2 0.2 0.2 0.2 1 1 17 Profitable 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 1 1 7 Efficient install 1 10 1 1 5 5 5 0.2 0.2 5 5 10 Efficient support capability 1 10 1 1 5 5 0.2 0.2 0.2 5 5 Performance (avg screen to screen & field to 2 field < 1sec) 5 10 1 1 5 5 5 5 1 5 10 3 Scalable (to 10,000 users) 5 10 1 1 5 5 5 5 1 5 10 18 Customizable 0.2 1 0.2 0.2 1 1 0.2 0.2 0.2 0.2 1 Lightweight on infrastructure (one measure 12 could be: works with Citrix) 1 5 1 0.2 1 1 0.2 0.2 0.1 0.1 1 Lightweight on in (one meas with C 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 46 FSS: easure 46
Excel HP Tool Sort by this column ascending to make the bar graph sorted Weight Factor 14 7.3 rchitected for next big thing 11 8.1 Customizable 6 8.3 Profitable 2 9.7 Time to market (Technology #2) 13 12.1 SP architected 16 21.5 Performance improvement for arket Segment #2 customers 5 22.9 Exceeds the competition 12 23.3 Lightweight on infrastructure (one measure could be: works with Citrix) 15 31.4 Performance improvement for arket Segment #1 customers 8 46.1 Efficient support capability 1 47.8 Time to market (Technology #1) 4 53.4 eets beta customer dates 7 58.9 Efficient install 17 75.4 Open architecture 18 75.4 Community connectivity 3 76 Clinically integrated 10 85.1 Scalable (to 10,000 users) 9 95.1 Performance (avg screen to screen & field to field < 1sec) 19 151 Product efect Containment 20 21 22 Use this column to "unsort" if you need to filldown for more factors 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 47 FSS: easure 47
Excel HP Tool Executive Priorities - HP results Product efect Containment Performance (avg screen to screen & field to field < 1sec) Scalable (to 10,000 users) Clinically integrated Community connectivity Open architecture Efficient install eets beta customer dates Time to market (Technology #1) Efficient support capability Performance improvement for arket Segment #1 customers Lightweight on infrastructure (one measure could be: works with Citrix) Exceeds the competition Performance improvement for arket Segment #2 customers SP architected Time to market (Technology #2) Profitable Notes: If results are not sorted go to the "Results (Sort Please)" Tab and sort as instructed. If you add factors update the range that is graphed Customizable rchitected for next big thing 0 20 40 60 80 100 120 140 160 HP Weight 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 48 FSS: easure 48
Prioritizing Requirements w/ Expert Choice T HP Start Expert Choice - Select Create new model (irect modeling method) Expert Choice software facilitates the data gathering and analysis for HP. We have permission to distribute their demo and tutorials, and we provide here a short orientation to their tool. Beyond that, we have no particular connection to the product. To learn more, or to get help while evaluating the demo software contact: James evlin P Professional Services 703-243-5595 escribe the Goal Tech Support 800-447-0506 www.expertchoice.com 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 49 FSS: easure 49
HP odeling Steps Right click select Insert Child of Current Node One builds a hierarchy by defining nodes under the goal, then adding child and sibling nodes to build out the detail. Type: OC <Enter> OB < Enter > < Enter > Important. an extra < Enter > closes out the node insertion User interface note: When typing in a series of nodes at the same level -hitting <enter> initiates a new node -you must terminate the process by hitting <enter> twice. Should look like this 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 50 FSS: easure 50
HP odeling Steps Open EC1.ahp simple hierarchy has been entered 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 51 FSS: easure 51
HP odeling Steps Synthesize (after assessments made) odel iew Pairwise ssessments * 1. Select Improving business results 2. Select BC above (*) 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 52 FSS: easure 52
HP odeling Steps Compare the choices with respect to the node s goal (Improving business results) Compare the relative importance with respect to OC\Improving business results The array displays results. Click in empty white cells to make specific judgments 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 53 FSS: easure 53
HP odeling Steps Note - Nodes with assessments to be done are marked - Completed nodes are solid - ssessments relate to the currently selected node one Not yet assessed Exercise: Complete all the assessments in the hierarchy - OC as an advocate for a customer you can represent - OB from your point of view 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 54 FSS: easure 54
HP odeling Steps When the assessments are complete, the indicators will change to boxes with green shading bars Select the node: OC Select Synthesize > With respect to current node 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 55 FSS: easure 55
HP odeling Steps Compare the choices with respect to the current node (OC) 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 56 FSS: easure 56
HP odeling Steps Synthesize all the choices Look also at Sensitivity Graphs to investigate how each factor weighed into the synthesis 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 57 FSS: easure 57
Going Further HP typically extends the use of this prioritization of criteria connected with the goal in two ways: 1. Synthesis of inputs from a group Each participant s evaluation of the pair-wise comparisons is stored and used in a group synthesis. Inconsistencies among the participants become part of the assessment 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 58 FSS: easure 58
Going Further dd alternative ssessment of the impact of alternative solutions, policies, scenarios, as courses of action to address the Goal. New pair-wise comparisons assess the impact of each alternative on the prioritized criteria. new synthesis around the priority of each alternative is then possible. We ll see an how that works in the next example 2003 Six Sigma dvantage, Inc. ll rights reserved FSS: easure - 59 FSS: easure 59