Data Science and Primavera Analytics...Project Success

January 25, 2019 | 2 minute read
Text Size 100%:

As mentioned in my previous blog post, the next series of post will be dedicated to getting into further detail on some of the recently released (Primavera Analytics 19.1) dashboards and analysis utilizing Oracle R and advanced analytics functions.

The first post will highlight the Project Success (Adv A) dashboard. This dashboard can be found in the Primavera Analytics 19.1 sample dashboards, under Dashboards, More, Project Success (Adv A) page.

For this advanced analytics function, our goal is to provide deeper insights given characteristics about a project; specifically if the project was successful or unsuccessful. We first determine project success (1 for success and 0 for unsuccessful) for all existing projects that are completed. Uncompleted project's characteristics are then analyzed and compared to characteristics of the already completed projects in order to find the most closely related project then use that projects successful or unsuccessful characteristic to predict and define our current uncompleted project as successful or unsuccessful. 

After all completed and uncompleted projects are assigned the new characteristic of successful or unsuccessful, we take a deeper look at the subset of projects that were categorized as unsuccessful. We decrease the numeric characteristics of each individual unsuccessful project by 5% for each iteration until the prediction determines the project is successful with those new decreased values or until the number of iterations has reached 10. If the number of iterations has reached 10 and the prediction still determines the project is unsuccessful, we keep the original numeric values for that project then move on to the next project.

Successful if:

  1. Completion Cost, Completion Units, Planned Cost, and Planned Units are all not null AND Completion Cost <= Planned Cost AND Completion Units <= Planned Units.
  2. Completion Cost and Planned Cost are not null AND Completion Units and Planned Units are null AND Completion Cost <= Planned Cost.
  3. Completion Units and Planned Units are not null AND Completion Cost and Planned Cost are null AND Completion Units <= Planned Units.

Unsuccessful if:

  1. All four values are null.
  2. All other cases - fail by default.

(NOTE: The above criteria can be changed as needed. These are merely meant to serve as an example of what is possible.)

In the next post, I will discuss the next part of Project Success, which involves making suggestions of how to turn an unsuccessful project into a successful one. If you get a chance, take a look at this new area in Primavera Analytics 19.1.



Shawn Lafferty

Previous Post

Five Feats of Digital Transformation: Oracle Aconex Connect Award Winners Show the Way

Janet Poses | 4 min read

Next Post

Owners and Developers Gain Control of Construction Payment Management

Corie Cheeseman | 3 min read