Promoting machine learning in the construction industry
Cloud providers such as Oracle are working to democratise machine learning and make it accessible to all players in the construction industry
From the perspective of the construction industry, machine learning (ML) has come a long way in terms of understanding what can be achieved using data science. However, there is still a long way to go before the adoption of processes such as predictive analytics becomes commonplace. In general, ML capabilities have increased significantly, especially in areas such as computer vision, use of neural networks and deep learning.
There have also been improvements in the ability to process big data using distributed computing, as well as the emergence of Machine Learning as a Service (MLaaS) from all leading cloud providers to democratise artificial intelligence (AI).
Additionally, there have been significant advances in existing open-source frameworks as well as new frameworks, which means the next five years could see a large uptake of ML technologies by the construction industry.
Data is the lifeblood for any AI and ML strategy. Many construction companies already have data available to them without realising it. This data, sourced from completed projects, can be used to build and train ML models, after which they can be compared against a validation test before being used for real-world prediction scenarios.
The accuracy of models can progressively improve by using data from different systems and sources, a scenario that is becoming increasingly possible with the digitisation of many aspects of the construction process. Often termed feature selection, the vast amount of data collected from different systems allows identification of the right data points that can help with improving model accuracy.
Furthermore, developments in automated machine learning (AutoML) is accelerating the model build process and reducing the trial and error that is often involved in improving model accuracy. Simplification of many of these aspects is also supporting the rise of citizen data scientists who are able to do more with the data than was possible before.
There are many emerging cases of ML being used in the industry. The most valuable ones are the ideas that will positively affect the metrics the industry has always valued – schedule, budget, quality, safety and risk.
For example, computer vision is now being used to identify progress on site, track delivery of materials, and, more recently, apply physical distancing rules on worksites. In general, ML techniques are being applied to unstructured data stemming from videos and photos to solve problems that were previously tricky to handle.
Natural-language processing (NLP) is now being deployed, which should reduce manual errors, improve productivity and mitigate risks. For example, NLP is being used to track submissions required for different jobs, identify non-standard terms in a contract, highlight potential safety hazards, or escalate the risk of an upcoming variation request. This will allow contractors and owners to better respond to problems.
The use of ML to predict schedule delays and cost overruns is another valuable contribution. These predictions will become better over time as data from different systems are integrated and fed into the models as inputs.
There is also a lot of smartification taking place. Schedules and budgets are becoming smart by incorporating ML-driven recommendations, supply chain selection is becoming smart by using data across different systems and comparing performances, and risk planning is getting smart by using ML to identify threats early and proactively.
Oracle Construction and Engineering is concentrating on what machines excel at when compared with humans and that centres around the ability to separate signals from noise in large amounts of non-visual data. Oracle is currently focused on building ML-powered predictive insights using Aconex, Primavera Cloud, Textura and Primavera Unifier data. Some of these use-cases will include:
- Predicting the probability of schedule and activity delays
- Identifying early warnings of underlying risks in activities within the plan, design and build phases
- Predicting budget performance and sources of potential cost over runs
- Forecasting variations and change requests ahead of time so that owners and contractors can manage their impact proactively
As it stands, the industry is only beginning to scratch the surface of ML-driven possibilities. As smartification drives datafication, a transformation driven by AI will naturally happen as companies begin to question how they leverage the data they possess. The cost of implementing ML is already decreasing, with infrastructure that enables pay-as-you-go models in the cloud. Oracle’s focus is on bringing the power of machine learning to as many capital projects as possible so that they can embrace purpose built data driven insights and intelligence out-of-the-box for strategic decision making without the need to invest in building significant internal ML capability which can often be quite daunting.
New tools are helping non-data scientists gain access to ML through drag-drop modelling, visualising predictions and simplifying the creation of easy-to-consume insights. Digitisation is set to disrupt every part of the supply chain and this is creating data that can change how built assets will be delivered in the future.
As processes are digitised in order to improve productivity, the data these systems generate will drive a paradigm shift in the process itself as machine learning will put to use the data that is created in ways that was not possible previously.
Karthik Venkatasubramanian is vice-president of data and analytics at Oracle Construction and Engineering.
Oracle Construction and Engineering, the global leader in construction management software and project portfolio management solutions, helps you connect your teams, processes, and data across the project and asset lifecycle. Drive efficiency and control in project delivery with proven solutions for project controls, construction scheduling, portfolio management, BIM/CDE, construction payment management, and more.