You have data – here’s what to do with it, says Karthik Venkatasubramanian, vice-president of data and analytics at Oracle Construction & Engineering
Data continues to fascinate the construction industry. Where other industries are much more mature in terms of their ‘big data’ journey, in construction it’s still gaining traction. Construction is often considered being behind in its digitisation journey, but in this weakness lies a significant strength: the ability to innovate like never before. And the opportunities presented by technology are starting to justify the cost more than ever.
We’re at a stage where it’s challenging to understand how construction and engineering companies managed without what we now consider ‘big data,’ particularly considering the level of control, transparency, threat awareness, and accountability it provides. Today, the industry appears data-hungry with companies keen to know what data science – including artificial intelligence (AI) or machine learning (ML) – can do for them.
This clamour for innovative data solutions invites some obvious questions: have construction businesses ignored the data they already have? How can they capitalise on the sheer volume of data they want? Can they even cope with more data? But also, is more data really the solution?
Data blindness
Before AI, ML and even the Building Information Modelling (BIM) methodology became widely understood, projects were still being delivered and quite successfully; just look at what the Egyptians, Mesopotamians or Aztecs achieved.
Many construction businesses already have data available to them without realising it. This data, arising from previous projects and activities, and collected over a number of years, might be sitting in rudimentary systems or spreadsheets. This often is not seen as “big data” as it doesn’t require huge expense or the integration of disparate systems through business intelligence solutions/AI platforms to generate insights. But in our experience, using data from any sort of system can help a business to leap ahead of where they were. It might not be integrated, clean or perfect, but there’s still value in it. The real trick is in finding out how to extract that value.
Capitalising on data
The art of deriving value from data often starts as a three-step process:
- Define the goals and frame the questions: Being clear about goals and questions is absolutely vital, ie are you trying to improve process turnaround, are you trying to better schedule performance? Is cost-blowout a specific concern or even a risk? Each of these questions will need to be tackled on its own based on the underlying source systems and data collected over time.
- Identify the data and analyse: For example, if you are trying to answer scheduling questions then you need raw scheduling information. This will help to answer questions around the quality of the schedule such as how accurate it was, whether the accuracy of the duration was correct, whether the schedule needed to be reconfigured. At the end of this analysis process, you’ve gained hindsight for future use.
- Apply the analysis to current projects: This is vital as it enables change to happen. For example, if the analysis reveals that hanging drywalls for certain types of projects by a specific subcontractor has typically taken 30% more time than allocated then the next project should change either the time allocated, how the drywalls are hung or the subcontractor employed. Using hindsight to gain insight that drives change is where the value of using data lies.
What about AI, ML and predictive?
Once the data has been collected, analysed and applied, it can be used as a training dataset for trying to predict future outcomes. For example, an AI-driven algorithm can automatically predict schedule delays in hanging drywalls on new projects using historical data and creating models that are based on a set of features (actual duration, project type, historical activity variance, etc).
The accuracy of the prediction gets better over time through learning from user input and new data that becomes available. This is where hindsight leads to the creation of insights, where models are built to gain foresight. Each of these steps can be independent and you don’t need ML to generate and use insights – any analytics tool can do the job.
What about augmented reality, virtual reality and Internet of Things?
There is significant value in these technologies to provide real-time, deep level knowledge about a project, both onsite and offsite. However, it’s important to understand that they would have to be used on a whole project before the value of the data captured could be recognised. Otherwise, there isn’t a complete picture to compare or contrast. There needs to be a baseline established.
The integration of these new technologies is not about creating isolated data sources. They need to be integrated and contextualised into an organisation’s current data flow. New technology innovations add to the data pool that is already available, enabling an organisation to gather richer, deeper insights. Eventually, all of this data will feed into gaining greater insights and driving better predictive outcomes.
Where does the data journey begin?
For many construction businesses, the journey has already begun. To see the value in a data approach, they need to explore what data they have and identify what they need it to tell them by creating a strategy.
Before it considers AR, VR, IoT or sensors, a business should ask itself whether it can even use the data it currently has. If not, it’s just adding to the complex pool of information that it already possesses. Either way, the key is to start from within, see what you have and grow your data approach from there. As with any large, complex undertaking, your data strategy needs a solid starting point. The insights will follow.
Karthik Venkatasubramanian
Vice-president of data and analytics
Oracle Construction & Engineering
+44 (0)207 5626 827
