Skip to main content

Interactive example using PERT estimating technique

From Beta distribution to PERT model estimating technique

http://brunociano.blogspot.it/
Beta Distribution
Regardless of the technique you use for estimating tasks, each estimate has to be quantified into a number. There are different ways for estimating project tasks, for example: 
  • Use of Historical Data - "similar" projects categorized in a knowledge database system can be retrieved and consulted when making estimates.
  • Use of Experienced Resources - estimating is a team effort, and the contribution from experienced resources in making estimation is very valuable.
  • Use of Estimating Databases - make use of industry databases made available by organizations.
In many cases estimates can be made more accurate by applying a simple PERT (Program Evaluation and Review Technique) model. PERT is an estimating technique that uses a weighted average of three numbers to come up with a final estimate.
The PERT distribution comes out of the need to describe the uncertainty in tasks during the development of a complex project having thousands of tasks. Estimates were needed to be made intuitively, quick and consistent in approach. The PERT distribution is a probabilistic model based on Beta Distribution, and it derives its estimates based on the probability of occurrence. A version of this four-parameter Beta distribution is called a PERT distribution and makes the assumption that the MEAN = (MINIMUM + 4*MOST LIKELY + MAXIMUM) / 6. The default value of 4 scales the height of the distribution. This extra equation allows the four parameters to be determined from three input values: the minimum, most likely and maximum, which makes it ideal for modeling expert opinion of  the uncertainty of a variable.
  • E = Expected Value = (O+4M+P)/6 (this is the "weighted" average equation
  • P: Pessimistic value (this value is maximum, because it describes the case when things go wrong)
  • O: Optimistic value (this value is minimum, because it describes the case where things go right)
  • M: Most Likely value (this value describes the case given normal problems and opportunities)
For example, let's suppose we want a more accurate estimate of a task duration that is collocated in a Project Network Diagram, given the following values:
  • P = 20 programmer days  (pessimistic value when everything goes wrong)
  • M = 15 programmer days (most likely) 
  • O = 10 programmer days (optimistic value where everything goes right)
Use the following interactive form to get the expected task duration (mean value) for the above example.
(Similar tasks can be categorized to create a histogram and it's possible to use the last chart below to highlight the mean value graphically, as stepped area across all other values).


Input Data (M,O,P)





Output Data Table (PERT Estimate)
(refresh rate every 10 seconds)






Output Data Graphically (PERT Estimate)
(refresh rate every 10 seconds)


Popular posts from this blog

The differences between Project, Operation and Program

We said that a project is defined as a temporary endeavor that consumes resources, incurs cost and produce deliverables over a finite period of time to achieve a specific goal. They come in all shapes and sizes and can vary in length or complexity.

Operation type activities are similar to project activities in that they too produce deliverables, consume resources and incur cost. However they are on-going or repetitive in nature, hence they are not project activities or tasks. Some examples of operation activities are weekly maintenance of databases, paying invoices or help desk operations activities.

Programs are much larger than projects. They are made up of many projects and on going activities such as operation type activities and are similar to projects as they consume resources, incur cost and produce deliverables. However programs are more complex and include repetitive operation type activities such as maintenance work, facility administration etc, and are funded typically on a…

Forecasting Project Costs using Variance Analysis

One way to report on cost control and forecasting during project execution is to use the Variance Analysis method, that is, explaining the difference (or variance) between actual costs and the budgeted costs with numbers and make new estimates for completing the work. Please consult this link Earned Value Management for related literature and references.
For the purpose of making these calculations, I will use an hypothetical project example (but it could also be a task or phase). "A company has contracted a service provider to deliver a project in 10 working days (80 hours) for the estimated cost of $10,000 and a work effort of 200 hours. The contract is Time and Material, this means that the company pays the provider for the number of hours actually required to perform the service. So, the provider has no incentive to minimize the number of hours expended on the service. The less efficient the provider is, the more money it makes!"
Summary of Time and Material Contract (re…

Using PERT for estimating tasks

A simple way for estimating tasks is to use the PERT (Program Evaluation review technique) weighted average method. This method uses a weighted average duration estimate to calculate task duration, it gives the opportunity to take into account information based on different types of estimates values (such as poorly defined areas, probabilistics, and ranges for the schedule). This method is based on the Beta distribution model because it can model events which are constrained to take place within an interval defined by a minimum and maximum value. (For this reason, the Beta distribution is used extensively in PERT, CPM and other project planning/control systems to describe the time to completion of a task).


The term weighted average means that the equation uses weighted factors to calculate the expected task duration.
The equation and process modelling a task for PERT is the following:  E=(O+4M+P)/6 (equation)  E= Expected Value  O= Optimistic Value (this is equivalent to a minimum val…