We are confounded by the terminology and the buzz around data science. It is a lot like looking through a dirty glass out into a hazy horizon to recognise what the colourful thing that moves out there is.
We get really frustrated when we can’t make out what it is. That is what it feels like when you enter the elite club of contemporary data scientists. Even though I wanted to discuss the aspect of contemporary data science challenges in Aviation, it is perhaps worth exploring what contemporary data science is.
Data Science is a practice as old as human cognition. YuvalHarari fans, who have read “Sapiens”, will understand what I mean. Everything we learn, we do by observation and to some level, a mental theorisation of the same. This means that our brain processes data and understands / interprets from it. We do the same (albeit at a much smaller scale) using computers.
So why is it that data science gets this exponential hype now more so than ever before? This is a question worth pondering over if you aspire to be a contemporary data scientist.
Companies have traditionally employed data scientists for a long time. Fundamentally, they are statisticianswho analyse, say for instance, market data and consumer surveys for example.
The typical questions posed to them could vary from: “Does this data suggest any significant improvement in our market strategy?” or “are all our consumers happy about the way we support them?”.
Measuring whether a strategy significantly performed better than the other or estimating the satisfaction of consumers is a simple enough problem for a statistician. I don’t insinuate that the traditional data scientists have an easy job. That is not at all true, but they have a clear defined problem to solve. There is so much data which is not curated that the keepers themselves have no idea what it might represent. This changed the textbook landscape of data science.
Data science has become pure Math at times, where you advance and then derive a use-case. The demands that we face now are questions to the nature of: “Take this search log data and try to find something useful we can use to our advantage” or “can you locate what went wrong with our customer strategy from the sales data for the last 10 years?”.
Let me compare the aspects of traditional and contemporary data science using the table below.
Now that we understand the flavours of contemporary data science, let us take a closer look at the domain: Aviation.
Let us take an example of an Aircraft or Engine Maintenance MRO that is trying to maximize the utilization of its capacity to optimize costs and increase revenue. This falls under the broad category of a finite capacity model: in other words, without increasing manpower, material, tooling, how can an MRO cater to more slots using the existing capacity – one that translates to greater revenue and optimize costs.
Historic maintenance data can help surmise the tasks and resource allocation model that has been most performant for various types of work scope levels against different Engine and Aircraft profiles. Such insights can be structured to possibility models which can simulate multiple shop / hangars visit schedules that give optimum performance towards maximizing profitability.
The simulation is done on many control parameters (Technical, Operational) which the planning process employs. This will allow us to benchmark a reference plan from the stage of a potential slot sale all the way until WIP execution: one that allows advanced prediction of profitability and risk mitigation.
One of our objectives at KeepFlying® is to expand the use of data science in interpreting Airworthiness and Maintenance data to generate commercial insights.
Flip the tables now: as an Asset Owner / Lessor, what use cases can you address? We will cover this as part of the next blog.