Cost Estimation:  Creating Proper Cost Models

Cost models, based on historical data, are great tools to aid high-level factor estimates. Cost models are computer models which take as input the relevant cost drivers of an object (such as a piece of equipment or an entire project), perform a calculation and return a cost estimation for that equipment or project.

When working with computers it is always important to consider the validity of the information entered into the computer: garbage in is garbage out. As a result, the same applies when developing cost models: when a model is based on wrong information, the model will return wrong costs.

Therefore, when collecting historical data with the intent of creating a cost model (or writing a project close-out report) it is important that the collected data is complete and consistent. All parameters which can influence the cost should be included in this quote collection overview. When in doubt collect as much information as possible, but do not lose track of what is important. Irrelevant data can always be excluded but when key data is missing the entire collection step will have to be redone. 

Tip: Try to structure or format your data such that it is obvious which parameters are likely to be the actual cost drivers and which likely do not have a big influence. An introductory text can also help both you and your (future) colleagues when revisiting the data.


Once you have a proper collection, a regression analysis tool - for example the direct cost trending tool in Cleopatra Enterprise - can be used to study the effect of the relevant cost drivers and develop a cost model. Study the regression analysis carefully to determine whether the model behaviour is accurate enough and carefully check the model after creation.

Keep in mind the desired accuracy of the model and for which stage of a project you want to develop the model for. For example, some information obtained from detailed engineering is not yet known during conceptual or basic engineering. Therefore, if you want to use this detailed information as an input for a Class 4 model, you will run into problems using the cost model because the information is simply not yet known.

To conclude

When developing cost models it is important to consider which information is available at the stage you want to apply the model, what the cost drivers are and whether or not additional parameters should be included to increase the accuracy. 

As a rule of thumb, try to reduce your model to the most relevant cost drivers; when the desired accuracy is not obtained investigate the inclusion of additional parameters.



Something to add to this article? Do you have a question for us, mail us at