Attributed to Karthik Venkatasubramanian, vice president of data and analytics at Oracle Construction and Engineering
Data provides the engineering and construction industry (E&C) with an opportunity to successfully manage built asset projects, using artificial intelligence (AI) and machine learning (ML) to predict different elements of the construction process.
To be successful, we need a data map showing the project now and where we want it to go with the data available. Here are four key phases to consider when building a data map. But even following these leaves the challenge of accepting what the data tells us, we’ll get to that later.
Formulating a data map
- ‘Flan’ (flexible plan): Define the problem and leverage data
Create a flexible plan to address what business problems need to be solved, what data is available, where the data exists, and how to use it. Define the analytics journey in a way that makes sense. This starts with learning from past data, applying past data to current projects and pursuits, and using this collective knowledge to predict the future.
- Learn: Accessing historical project data
Mining historical data for insights to find answers to questions including: How long do processes take on average? Who are the best and worst performing sub-contractors? What activities have typically been delayed in the past? These help to define baselines, benchmarks, key performance indicators (KPIs), and standards. Historical data offers an excellent starting point.
- Apply: Reviewing historical insights to current projects
Applying historical insights to current projects to answer questions like: Are the estimates built right? What did we learn from the past? Have we constructed the schedule correctly? Are these the best partners to work with?
- Predict: Schedule, budget, quality, safety, and risk
Use the data collected so far along with ML to predict future outcomes about things the industry cares about: schedule, budget, quality, safety, and risk. Complementing lag indicators with lead indicators including understanding the probability of delay on projects, amount of predicted delay, likelihood (and severity) of a cost blow-out, and hidden safety, design, rework, and litigation risks within a project can transform the business.
Creating a data plan is the easy part, the bigger hurdle the industry needs to overcome is trusting what the data tell us about what might happen on a project – the prediction.
The challenge of probability
We all like certainty but the industry needs to start accepting probability in order to get the most out of the data available. Certainty tells us about what has or is happening on a project (known as a lag indicator). To look at what might happen (the lead indicator) we need to appreciate the value of probability.
It’s one thing to collect and store the data, it’s another to take the step towards trusting what the data can tell us about the future of the project. This is even more so when the outputs of ML are probabilistic in nature.
The key word is trust. Making sense of the data and converting it to insights and intelligence is best left to the experts. Realistically, you want to make data available to project participants, but you don’t want everyone to be a data expert. You need specialists to ensure certainty of results and reduce risk of errors. This goes someway to instilling that trust.
Trust also comes from accepting the way we use probability in other areas of our lives and transferring that to projects, such as our acceptance of video streaming recommendations or weather predictions. While most of us use them regularly, we don’t really know how the predictions are made, how accurate the predictions are and what data went into making them. Yet, we generally trust them.
The same principle applies to fraud detection, sales projections and several other areas sometimes labelled as ‘number crunching’. Typically, we have accepted them even though they are inherently probabilistic. We need to do the same with data in project and construction management.
This is not a linear process and the journey in some ways is all about the initial steps. The technology is shifting, the industry is changing, processes are transforming, and the next generation is entering the industry. It’s not about sprinting to the end, but about starting the journey of trusting probability today.