Passenger satisfaction is currently one of the most significant key performance indicators (KPIs) for benchmarking airports and airlines. It also yields valuable insights on airport processes, including their effectiveness and efficiency. Not only does it provide understanding and awareness of airport, passenger and airline processes but also offers detailed analysis, trends, and pattern recognition in retail revenue development and marketing initiatives based on personal preferences.

In the context of the biggest Horizon 2020 EU-funded data analytics research project for the transport sector, the TransformingTransport project brought Athens International Airport together with Indra, Spain, to approach airport data analytics describe relevant use cases.

Analysing passenger behaviour, segmenting passenger traffic in categories, identifying demand for airport facilities and forecasting passenger numbers not only helps outline the passenger process but also offers strategies that will significantly increase passenger throughput and enhance passenger experience.

Our first steps in the analysis

Indra put together all sources available (figure 1) to develop descriptive and predictive models in order to forecast airport passenger volume and airport resource demand. This prediction foresees possible bottlenecks and strives to reconcile airport resources with forecasted demand. The more successful the process is, the greater the airport efficiency and passenger satisfaction.

Win-win collaboration with all the stakeholders: the key to success

A win-win partnership was established between airport operations, airline home carrier (Aegean Airlines) and ground handling companies. These synergies led to sharing experiences and insights on passenger operations, including the interface of processes between airports, airlines and ground handlers. This also involved data exchange, focusing on what customers need in order to enhance business models.

Use cases and Key Performance Indicators

One of the most important aims is to improve maximisation of passenger flow and passenger experience while reducing aircraft turnaround delays caused by passengers such as no-shows at the gate.

To achieve this, Indra teamed up with Athens International Airport operations professionals to identify Transforming Transport’s main KPIs. These include time distribution of check-in desk arrivals, time distribution of security control arrival, dwell time and check in channel.

Technology and techniques applied in the analysis

Data mining or exploitation of information is a process to extract useful, comprehensive and new knowledge mostly from large data volumes to find hidden or implicit information. Data mining is implicit in Big Data analysis and has been the main technology that helped develop the project successfully. The main steps are represented in Figure 2 and can be resumed in three main and important steps: data preparation, creation of advanced models, and results validation through machine learning algorithms.

 

One of the last and more important steps is the interpretation, review and adjustment of the data results, along with a deep business analysis that better addresses business goals.

Results of Data analysis

Applying all the above in several use cases has yielded the following results:

• Optimisation of transfer passenger arrival to the security screening area: to optimise passenger flow, the airport/airline should provide incentives to passengers to proceed to the security screening points as early as possible for a more even distribution of passengers.
• Distribution of dwell time for transfer passenger by waiting time: passengers tend to stay at the airport when the time difference between the incoming and the outgoing flight is six hours or less.
• Clustering: economy passenger on domestic/international flights: the majority of passengers on international flights arrive 15 minutes earlier than those on domestic flights.
• Retail sales per passenger segment of specific flights: the retail spending behaviour/performance of each flight passenger segment can be mined so that best performing flights and segments can be identified. Those insights can be extended even further with retail basket analysis data that can provide information on passenger retail preferences to be used for personalised offers, thus enhancing passenger satisfaction.
• Retail sales based on flight allocation at different parking stand categories: identifying the spending patterns of passengers per flight and assigning those flights close to the airport’s retail area maximises retail revenues for retailers and airports. This is an example of how the airport can maximise retail sales through airport operations.

Conclusions

Data analysis needs several data inputs, both historical and real-time operational ones. By developing the use cases and relevant KPIs, we successfully predicted flow of passengers arriving at the security areas and optimised security personnel. At the same time, we ensured the appropriate service levels required to enhance passenger satisfaction.

Data analytics facilitates the analysis of elements that affect terminal operations through descriptive models that analyse the current situation and predictive models that predict future operations.

This led us to identify current passenger needs and gain insight on passenger flows, behaviour and segments. It provided new input for improving the terminal design process and current use of airport terminal resources.

Regarding retail, passenger segmentation and clustering has enabled us to identify the different passenger behaviour trends which will lead in the future to personalised services for improving passenger experience and sales.