In a construction business, the importance of the accuracy and the precision of the cost estimates of projects cannot be overstated. Overestimates causes one’s bid price to rise, resulting in the loss of opportunities. Underestimates keep one’s margin below what the market can support. Needless to say, it is highly desirable to have estimates as close as possible to the actual cost.
In order to improve the accuracy and the precision of one’s estimates, it is important to find out what is causing them to be less so. It could either be the factors that influence the estimates or the factors that influence the actual cost. It could be a combination of multiple factors. A visual analytics tool that can help us explore and find answers is a box-and-whisker plot. One such example plot is shown below (an interactive version of it can found at the end of this post).
In this plot, we analyze the cost variation as a function of estimators. The x-axis shows the variation of the actual cost (actual cost – estimated cost) as a percentage of the estimated cost. Each dot represents a project. The plot on the left shows the variation for all the projects. The plots on the right show the variation for the projects grouped by their estimators. The grayed area of the box shows the spread of the 75% of the projects. The boundary between the darker and lighter shade falls on the median. The dotted lines show the average variation for all the projects.
For better precision, we would like the spread to be as narrow as possible and for better accuracy we want average and median to be as close to 0% as possible. However, from the plot, we can see that the accuracy and precision vary significantly from one estimator to another. For instance, in the example shown, F. Cage has the most precise estimations but their accuracy on average tend to be off by 30%. The estimates by C. Mancino, on average, are the most accurate (18% off) and they are also more precise than that of other estimators. On the contrary, P. Parish’s estimates seem to be the most imprecise and inaccurate.
The cost variations may also be attributed to factors other than estimators and it would be important to analyze the impact of those attributes as well. The chart below shows variation analysis by project managers. In this example, E. Zegar’s projects have very wide cost variations and are not very accurate while D. Musich’s projects have actual cost that varies much less than the original estimates.
Often, it would be important to keep one attribute (such as an estimator) fixed and compare how another attribute (such as project managers) contribute to the variability. An interactive version of this chart (provided below at the end of this post) allows one to perform such an analysis. For example, by selecting an estimator (for example C. Mancino) in the first chart, we can see cost variability of those projects by project managers
Other similar analysis may be performed for others attributes. The client, division or the type of projects may be the attributes that are causing large variability. In your business, what do you think such attributes are and how influential are they? If you don’t know, doing this kind of analysis can certainly help identify them and their influence.