Food for thought: can predictive analytics make the airline industry more sustainable?
4 min.
Using cutting-edge data science technologies, such as artificial intelligence (AI) and machine learning (ML), LACO brings a modern, innovative and, above all, sustainable business approach to the transport industry.
What’s the first thing you think of when the environmental impact of flying is mentioned? Fuel consumption? Obviously that is a hugely important factor to take into consideration, if the ecological footprint of the airline industry is to be further reduced in the coming years.
Research shows that airlines generate lots of waste.
6.1 millions tons of cabin waste in 2018.
Source: International Air Transport Association
However there are other factors at play. Less obvious maybe, but not necessarily less relevant. Take food waste, for example. Did you know that no less than 6.1 million tons of on-board meals were discarded by airline companies in 2018? The impact of that waste is bigger than you might think. Why? Because it means that next to spoiling food, an enormous amount of fuel was wasted to take these unnecessary 6.1 million tons of food through the sky. With a better forecast of on-board meals, airlines would save lots of money on food… and fuel.
Rise to the challenge
Last year a LACO team tackled this somewhat neglected issue in a hackathon competition aimed at improving sustainability through technological innovation and, more particularly, through data analysis and machine learning. The hackathon required the presentation of an innovative use case connected to at least one of the United Nations’ Sustainable Development Goals (SDG) .
Our case, dealing specifically with the use of predictive analytics for sustainable transport and, on a broader level, climate change, came in second in the European series of that competition. And obtained the first place amongst the Belgian applicants. An achievement that once again confirms our leadership position in the field of data analytics in Belgium – and beyond.
Predictive Analysis in Transportation
The solution we developed is based on two elements:
1. PAT – Predictive Analytics in Transportation
First of all, we developed a web application called PAT (Predictive Analysis in Transportation), based on NodeJS and using the SASJS library to trigger the machine learning engine on the SAS platform. This web application displays the result of any prediction you want to trigger.
2. SAS platform
The second element is the SAS platform itself, used to clean and transform the data before running a ML algorithm on the model. Based on the result of the comparison we made, we decided to opt for a Gradient Boosting Tree algorithm.
Watch this video for a quick presentation of our use case a more detailed view on the project:
Combining the best of two worlds
Organised by SAS , one of our strategic technology partners, the hackathon competition offered another opportunity to showcase our expertise with this vendor’s analytics platform. But what made this hackathon even more interesting to us, is that we could combine our SAS skills with our knowledge of Microsoft cloud technology. SAS Viya on Azure is the cloud analytics solution in question. And we were all very impressed by its ease of us, which made for an easy deployment process and a rapid delivery of concrete results.
When it comes to implementing SAS on Azure, there is a clear reason why we got to be the first SAS partner to do it at a customer site in Belgium . More than one reason, in fact, we have one of the strongest SAS teams on the Belgian market and on top of that Microsoft is a strategic technology partner. In other words: we firmly have our finger on the pulse of two leading vendors in the business intelligence space.
Cherry on the cake
But let’s get back to our sustainability showcase. Basically, we started out by trying to answer the following simple question:
Can analytics help to predict the exact number of every meal option sold on a particular flight?
With PAT, we offer a solution that allows a non-technical user to run a prediction based on basic flight criteria. This way PAT may well be the key factor in bringing advanced technologies for predictive analysis to every non-technical employee in your company. On the other hand, that same interface is just the tip of the iceberg, so to speak. Or better yet: the cherry on the cake.
Check out this longer (15min) video presentation to find out all the details about the underlying advanced data science technologies that made our use case for the airline industry such a success at last year’s SAS hackathon:
The promise of AI and ML
The on-board meal waste was reduced by 80%
Our project proved that ML could pave the way to a strong reduction of food waste in the airline industry. Based on our model, the waste of on-board meals can be reduced by an astonishing percentage of 80% and as such resulting in lower fuel consumption. Adding more data – such as passenger data and weather forecast data – will only lead to better and more accurate results. Moreover, linking the SAS Engine to the airline’s flight planning system would allow to automate the calculation of the on-board meal prediction.
And there’s more. The same solution could be used to predict airplane maintenance, fuel consumption, and other processes in the airline business. And why not broaden it even more, and predict on-board meals in trains or on cruise ships? All in all, our user-friendly solution brings AI and ML within reach of any transportation industry, offering a modern approach of running its business in a more sustainable way.
If you want to discover the opportunities of AI and ML for your business? Or talk about the benefits of deploying your SAS environment on Azure? We’re happy to have an open discussion!
Benjamin Pierre
Junior SAS Developer at LACO
