4squareviews

In the last part of this two-part post, I went over the tools and techniques for this process 6.4 Estimate Activity Durations which I call generic, because they are used for any complex planning process, and not just the ones for schedule management.   These are

• Expert judgment
• Decision Making
• Meetings

In this post, I will cover the tools and techniques that are used specifically for this particular process 6.4 Estimating Activity Durations, namely:

• Analogous estimating
• Parametric estimating
• Three-point estimating
• Bottom-up estimating
• Data analysis (alternatives analysis, reserve analysis)

Let’s discuss these tools and techniques (the numbering is based on the categories in the PMBOK guide, so there will be a little skipping of  numbers–this is intentional and it means that the missing categories are the ones covered in the last post on the “generic” tools and techniques for this process).

6.4.2  Estimate Activity Durations:   Tools and Techniques

6.4.2.2  Analogous Estimating

6.4.2.3  Parametric Estimating

I am going to discuss these two techniques together because they are both examples of top-down estimating techniques, that is, estimates based on the total cost of previous, similar projects.   They both require historical information on similar projects in order to work.

Let’s illustrate these techniques with an example.   Suppose I work for a construction company, and we bid for a project to build a new home in an already existing subdivision.   How do we estimate the amount of time it will take to complete the home?  Well, one quick way is to see how long it took our company, or perhaps another company or comparable size, to complete a similar home in the subdivision.    If it took then 90 days to build it, we can use the analogous estimating technique to calculate that 90 days is a good rough estimate for our project as well.

However, say there are no homes that are comparable to the one we are doing in size.   Maybe there are small and medium-sized homes, but not one that is as large as the one we want to build.   In that case, a more nuanced estimating technique called parametric estimating can be used.    If you have data on how long it took to build homes of various sizes in the same subdivision, and you have how large these houses were in terms of square feet of floor space, then you can calculate a parameter or unit measurement which estimates how long it takes per square foot of floor space.

If the average home in our subdivision has 2,700 square feet of floor space, and we assume it takes 90 days to build it, then it takes 1 day per 2,700/90 = 30 square feet of floor space to complete an average home.    If our project is to build a house that is larger, say, 3,000 square feet, we can make a rough estimate that it will take 3,000 square feet x (1 day/30 square feet) = 100 days to complete the home.

This is a rough estimate, and it is based on assumptions such as

• the other homes being built are comparable in terms of floor plan
• the building materials being used are similar
• the experience level of the companies that did the other houses is comparable to that of our company

So the advantage of these techniques is that they are quick, but the disadvantage is that they are rough estimates are therefore not as accurate as a bottom-up estimating technique (see paragraph 6.4.2.5 below).

6.6.2.5  Bottom-Up Estimating

This is when you estimate the duration of each activity on the activity list, and then aggregate the estimates for the differing level of components in the Work Breakdown Structure or WBS.   First you sum up the durations of each activity for each work package, then you sum up the subtotals for each work package, rolling up the levels of the WBS until you get the total estimate for the entire project.

The advantage of this technique is that it is accurate, but the disadvantage is that it is time-consuming.    It is the exact inverse of the advantages and disadvantages of top-down estimating techniques (discussed above in paragraphs 6.6.2.2 Analogous Estimating and 6.6.2.3 Parametric Estimating).

6.6.2.4  Three-Point Estimating

I am putting this below the bottom-up estimating technique because that is where you start, with a single-point duration estimate for each activity that is made in the bottom-up estimating technique listed above.

You further refine this one-point estimate with the three-point estimates listed below:

• Most likely (M)–this is based on the duration of the activity, usually the one found in the bottom-up estimating technique, based on the assumptions that will most likely occur such as:
  • the resources likely to be assigned to the activity
  • the productivity of these resources when performing the activity
  • the realistic expectations for the availability of these resources for the activity
  • the dependencies of the resources on the other participants in the project team, etc.
• Optimistic (O) –this is based on the best-case scenario for the activity
• Pessimistic (P)–this is based on the worst-case scenario for the activity

The expected duration can be calculated with one of the following formulas:

1.  Triangular distribution

E = (O + M + P)/ 3

in other words, the simple average of the three-point estimates.    This is done when there is not a lot of historical data regarding the activity.

If there is historical data regarding the activity, and you have more confidence in the “most likely” estimate, you can statistically give it more weight by using what is called a “beta estimate” (I remember this by thinking of it as a “betta” estimate than the triangular one).

E = (O + 4M + P)/ 6

In this case, you divide by 6 inside of by three because in essence you have copied the “most likely” estimate of M 4 times, and so you have a total of six terms to take the average of rather than just three terms that you have with a normal average.

Let’s see how this works with an example.

Okay, let’s say you move to a new house, and your boss asks you how long it will take you to get to work from your new place.   On a typical work day, you see the estimate of the time you leave your house to the time you get to your work place to be 30 minutes.   Not bad!    However, that is the “most likely” estimate, if traffic is normal and you only have to deal with “rush-hour” traffic.

What would be the “optimistic” and “pessimistic” estimates for getting to work.   In the optimistic or “best-case scenario”, there would be very little traffic.   This can happen if you are coming to work on a holiday or maybe even on a Saturday, when there is no “rush-hour” traffic to deal with.   Let’s say it only takes 20 minutes to get to work in that case.

How about the pessimistic or “worst-case scenario”?   That might be if there is a traffic accident which causes traffic congestion that is worse than the normal “rush-hour” traffic.   In that case, say it takes 60 minutes to get to work.

What is the three-point estimate of how long it will take to work?    If you are new to your new route to work, and don’t have a lot of historical data to compare it with, then you might want to go with a regular triangular estimate:

E = (O + M + P)/3 = (20 + 30 + 60)/3 = 40 minutes.

If you tell your boss it an estimate of 40 minutes, then since it most likely will take you 30 minutes, you will always make it to work on time if you leave 40 minutes before you are scheduled to work UNLESS you face the pessimistic scenario of an accident on the freeway or highway you are taking to work.

Let’s say you get some more confidence in your route to work, and you have never seen an accident yet on the road to work, so you can now use the beta or weighted average to calculate your time to work.   Now

E = (O + 4M + P)/3 = (20 + 4 X 30 + 60)/6 = 33.3 minutes.

If you give your boss an estimate of 33.3 minutes, then you now can tell yourself to leave 3.3 extra minutes rather than 10 and still be reasonably confident that you can make it to work on time.

That’s how three-point estimating works.   It takes the single-point estimates found in bottom-up estimating and creates an estimate that is robust, that is, will work under most scenarios but perhaps the most pessimistic.

6.4.3.6  Data Analysis

The techniques used to analyze the duration estimates of activities are:

• Alternatives analysis–this is an analysis of the assumptions behind the various options you have for doing an activity.   These assumptions come into play a lot when doing the three-point estimates described above.
• Reserve analysis–this has to do with determining contingency and management reserves needed for the project.   Contingency reserves are those that can be used by a project manager, and management reserves are those that need approval of management (such as a project sponsor).

Let’s take the example we talked about before.   It takes 30 minutes according to your GPS for you to get to work.   The three-point estimate shows that, in order to account for the possibility of some unforeseen circumstance that might cause you to take more time, a better estimate is 33.3 minutes.   This extra 3.3 minutes could be considered a contingency reserve which would allow you enough extra time if there was some sort of slowdown on the highway that caused you to take more time than you normally would (i.e., 30 minutes).   This is a reserve that you control yourself and it helps you get to work on time in most cases.

However, if there was some sort of traffic emergency on the road, you might have to call in to your boss and tell him or her that you might not make it by the time work starts.   The boss will usually say “okay” given the nature of the emergency, and this is like a management reserve which allows you to take the extra time you would need provided that is is an emergency.    “Hey, boss, sorry I forgot to set the alarm clock” would not be considered sufficient circumstances to allow you a “management reserve” of extra time in this case.

There will be more on these estimating techniques when we turn to the next knowledge area of schedule management.

With this discussion of tools and techniques for schedule management being concluded, let’s now turn in our next post to the outputs of this process.

Filed under: Uncategorized |