Artificial intelligence in construction projects: From fail and fix to predict and prevent

Construction projects requires diverse teams to plan, design, construct and uphold the project. It is commonly agreed that for a construction project to be successful it has to be completed on time within budget and according to the specification.

Forecasting project performance is one of the most demanding tasks in predicting whether the project will be a success. The effective performance of construction project cannot be achieved without challenges and obstacles. To meet these challenges and beat these obstacles, an organisation must have a clear awareness of its performance.

Efficiency in the construction can be defined as the project completed within the time schedule and cost budget. A project can therefore be defined as a series of unique, complex, and linked activities having one aim or purpose and that must be completed by a specific time, within budget, and according to requirement.

How do you measure project health?

Firstly, a construction project can be defined as a sequence of unique activities, having one goal or purpose and that must be completed by a specific time, within budget, and according to specification. Project performance is judged mainly based on the four performance metrics i.e., cost, schedule, quality, and satisfaction performance.

A good project manager will use experience and historical data to identify obstacles and warning signs in an ongoing project, to take corrective actions or even mitigate risks already when starting and planning the project.

The decision-making and management may be supported by live data dashboards, gathered from the project, displaying KPI´s and visualising project status to assist the manager in the decision-making process.

This is defined as descriptive analytics.

The future is predictive analytics or artificial intelligence / AI.

Algorithms and challenges

For implementing the correct algorithms for making predictive analysis, you need historical data, analysed and structured to help the programmers / mathematicians implementing the right algorithms.

Finding the right parameters, variable references, scaling and weighting these is a complex and tedious task, especially when working on different types of construction projects and with different users that might look at the results differently.

User friendliness is key

Machine learning and neural network algorithms are very complex, and not for you average John Doe to implement, but rather highly educated mathematicians and computer scientists. Hence, the presentation is very often complex and hard to interpret. Especially in construction, user-friendliness is important.

A survey from Software Connect, states that Ease of Use is the number 1 important consideration when buying construction management software. At Fonn, our first priority is making our system available for all users with an inclusive usability for all tech skill levels.

The solution

Fonn has been gathering data from more than 65,000 projects over the last few years, with high quality data, structured for the implementation of a neural network.

The predictor variables have now been selected and the historical “health” of projects analysed. Interviews and workshops have been held to set the parameters of cost, schedule, quality, and satisfaction performance and other critical factors on completed projects.
These outputs are now being used to help our users in their decision-making processes and focusing on the “unhealthy” projects instead of the projects that are performing good and moving the industry from Fail and Fix to Predict and Prevent.