## Friday, October 25, 2019

### Book Excerpt from "I Want To Be A RMP", 2nd Edition - Representations of Uncertainty

It is from Chapter – 8: Quantitative Risk Analysis.

For the partial index of the book, refer the above link for the book.

To know what is NEW in this book's 2nd edition, refer:
What's New - I Want To Be A RMP Book, Second Edition

For overall details of the book, refer:
https://www.managementyogi.com/p/books.html

*********

Representations of Uncertainty

The individual project risk arises from sources of uncertainties, e.g., as we just saw in probability distributions – you can reach you friend’s house in 1 hours (best case), 2 hours (most likely) or 3 hours (worst case).

In fact, there are many ways you can have such representations of uncertainties, including probability distribution. For your RMP exam, you need to be familiar with these three:
• Probability distribution
• Probabilistic branching
• Correlation

8.10.1 Probability Distribution
We have discussed probability distribution in-detail in the earlier section.

[Note: In this site, there is an exceprt on various probability distributions. More have been added to the 2nd edition of the book. Link: Probability Distribution in Risk Management]

8.10.2 Probabilistic Branching
This is another way to represent individual project risks. In this case, risks are included in the project model as probabilistic branches, i.e., you model the risk of different outcomes occurring in a project. Obviously, due to branching, we will have paths or branches with respective outcomes. In other words, you can say probabilistic branching is used in quantitative risk analysis when the outcomes are mutually exclusive, i.e., only one of the outcomes occurs.

However, to understand probabilistic branching, you need to first understand a concept called "Activity Existence". This is because branches in a project will be between two activities (tasks) or among multiple activities (tasks). Activity existence tells the probability that an activity can exist in a schedule network diagram.

When an activity's existence is probabilistic, i.e., it may not exist or may exist with a probability, then obviously, it will impact the duration (and cost) of the project. If an activity or task does not exist (0% chance) during a risk analysis, then there can't be any probabilistic branching between the task under consideration and another task.

Next, there can be two types of probabilistic branching.

8.10.2.1 Probabilistic branching, single branch
In this branching, the activity exists on a single branch. An example is shown in the below figure.

As shown in the above figure, we have a branch in a network diagram with 3 activities - A, B, and C, with durations of 3, 4 and 5 days, respectively. However, the activity B has a 30% chance of existence.  During quantitative risk analysis, this “Activity Existence” probability will be considered.

8.10.2.1 Probabilistic branching, multiple branches
In this type of branching, the activity exists on multiple branches. An example is shown in the below figure.

As shown in the above figure, we have four activities here, with activity A being succeeded by three other activities – B, C and D, with 20%, 35% and 45% chances, respectively, as their existence values. The durations for these activities are noted next to their names. Considering the above figure, we have 3 branches:
• Activity A followed by Activity B,
• Activity A followed by Activity C, and
• Activity A followed by Activity D.
These activities are modelled with their chances of existence before conducting risk analysis with simulation. For example, considering our previous probabilistic branching with multiple branches, the network diagram is shown below as a risk-adjusted Gantt chart. The image is drawn with Primavera Risk Analysis software.

As shown in the above figure, the activities in the above Gantt chart are risk adjusted, with the probabilities for existence shown for them - 20% for Activity B, 35% for Activity C, and 45% for Activity D.

Next, as we do risk analysis, the most likely or least likely branches will be chosen based on value that we enter for our simulation. Depending on the number of iterations and branches chosen, our project’s end date will be impacted. If you considering both the durations and cost for the activities, then not only the project’s end date, but the overall project cost will also be impacted.

So, probabilistic branching is useful when the outcomes are exclusive, i.e., the associated risks are not related. How about these situations when the outcomes are not exclusive? In other words, how about scenarios when the risks are related?  Will the probabilistic branching be used? No! In such cases, we use “Correlation” – another representation of uncertainty.

*********

This section in Qualitative Risk Analysis is further explained in the book with:
• Correlation for risks (coming from correlation of tasks/activities).
• Other data analysis techniques used such as Monte Carlo Simulation, Latin-Hypercube Simulation.
• Subsequently, we have discussions on Risk-adjusted S-curve analysis, Contingency reserve calculation etc.

It is further followed by detailed explanations on various aspects of quantitative risk analysis as outlined in the Book Index.