The aeronautics world is a complex web of interconnected systems, each with its own set of complexities and challenges. To effectively manage these systems, comprehensive understanding and control over data is crucial. This is why data has become an essential topic in modern aviation, aiming to maximize knowledge recollection on given features. However, one often overlooked question, is how much we can trust all the available information. This particular aspect is what we aim to address when focusing on data quality.
Unfortunately, data trust concerns cannot be answered by a simple yes or no. Different stakeholders within an airline’s data project might have different approaches to data quality. For instance, a pilot might focus on technical details of his flights' data, while a fuel efficiency manager might be concerned by the reliability of last week’s operations data.
This article will explore the importance of data quality in aviation, highlighting its direct influence on fuel efficiency, and exploring the best practices for maintaining high data standards.
What is the role of Data Quality in flight operations?
Data isn’t something we lack of when it comes to aircraft operations. Data is abundant. From flight planning and in-flight data systems, to all mandatory documentation, there is a wide spectrum of data sources available. While this wealth of information can be a powerful tool, it can also a double-edged sword. On the one hand, we have plenty of information to have a comprehensive view of a given flight. On the other hand, it means there are many places where inaccuracies or inconsistencies could occur.
💡 But what is the value of all this knowledge for airline operators?
Accurate data enables you to make precise decisions, that can have a positive impact on operational metrics your airline wants to improve.
This is where data quality becomes essential: anything distorting your analysis to different degrees could have a big impact on the results. Poor data quality can lead to misguided solutions, you could be solving a problem that doesn’t exist or, even worse, you could think your solution isn’t working when it actually is.
The Takeoff fuel consumption anomaly example
Analyzing fuel consumption data can sometimes reveal unexpected insights. For example, in this case, average fuel consumption during takeoff appeared to be surprisingly low. At first glance, this seemed like a great outcome, efficient fuel usage during takeoff is always a positive sign.
However, a deeper look uncovered something unusual: in several flights, the data showed fuel gains at the moment of takeoff. Since aircraft don’t generate fuel mid-air, this was clearly an error. The most likely explanation? The forces acting on the aircraft during takeoff were interfering with the fuel sensors, causing inaccurate readings.
As a result, these misleading data points had to be excluded to ensure a more accurate average fuel consumption figure, one that wasn’t artificially low due to faulty measurements. This highlights an important lesson in data analysis: sometimes, metrics that look good on the surface may not be telling the full story.
There are two key perspectives on data quality for fuel efficiency: the pilot’s and the IT team’s. While both rely on accurate data, their concerns and interpretations can differ significantly. To ensure they speak the same language and address the same issues, the topic has been structured into six core concepts, known as Data Quality Dimensions. These dimensions provide a clear framework for defining and assessing data reliability, helping bridge the gap between operational and technical viewpoints.
What is it ? Accuracy refers to the degree to which data factually represent its associated real-world object, event.
Example: If the recorded fuel level at arrival is 1,000 kg/2204 lbs but the actual amount is only 700 kg/1543 lbs, there’s a clear discrepancy.
Solution: Consolidation to improve accuracy. Multiple data sources should be cross-checked to determine the most plausible value. Over time, understanding which data sources are the most trustworthy helps establish a more accurate reference.
What is it ? Completeness refers to whether all required data is present.
Example: No fuel consumption data is recorded for the taxi phase of a given flight.
Solution: Implement clear dashboards to track the assimilation of different data types over time. Alert and inform when data is missing or when new data could provide additional insights, unlocking new possibilities.
What is it ? Consistency is the ability to correctly link data relating to the same entity across different datasets.
Example: Fuel recorded at departure is lower than fuel recorded at takeoff, an inconsistency that suggests a data error.
Solution: Apply consistency checks to flag flights presenting undesired behaviour. These flagged cases can then be excluded from specific analyses to prevent misleading conclusions.
What is it ? Timeliness measures whether the delay between when something happens (real value) and when the data is accessible is appropriate for its intended use.
Example: Flight data becomes available one month after the flight, delaying any potential analysis or decision-making.
Solution: Track data integration times and work on optimizing the overall process to ensure faster availability.
What is it ? Transparency refers to the extent to which data is well documented, verifiable, and easily attributed to a source.
Example: There’s a two-hour discrepancy in a flight’s recorded departure time, but the cause is unclear.
Solution: Enhance data lineage visibility by documenting where each data point originates, whether it is raw or has been transformed during processing. This helps contextualize discrepancies and improves trust in the data.
What is it ? Validity refers to whether data values are consistent with a defined domain of values.
Example: A pilot notices that the recorded approach for landing at JFK is one they would never use, suggesting the data is incorrect.
Solution: Empower the user. No matter how advanced the system is, experts on the field have firsthand knowledge of what actually happens. Maintaining open communication with users allows for smart adjustments in edge cases, improving overall data reliability.
Acting on enhancing any of these different dimensions is a never-ending process, very similar to security in aviation: multiplying initiatives will ultimately lead to low and accepted error rate.
-
Check at the source:
Most of the data issues encountered can be traced back to problems originating in the source file. When source problems aren’t corrected on time, it can propagate the issue, making it hard to track it back to its origin. So, ensure that data is verified and validated at the entry point to maintain overall data integrity. -
Track the issue:
Fixing an undesired issue is not as important as making sure it doesn’t come back! Indeed, preventing data inconsistencies from reoccurring is critical. To minimize these disruptions, set up a system that alerts your team when a known issue occurs. -
Spread the word:
The most effective system will always be the vigilant expert who spots inconsistencies. Identifying something that doesn’t make sense can have notable positive repercussions. So, encouraging team members to remain vigilant and report anomalies that don’t align with expected data patterns can lead to significant improvements.
To sum up
Maintaining high data quality in aviation is essential to having a solid base for all new cutting-edge data technologies. Often looked down upon, it will make a difference in the long run in having better metrics and optimizing fuel efficiency.
At OpenAirlines, everything starts when we get our hands on the data; we perform initial tests to validate that everything we get into our system is in the correct format. At each step of our process, we continuously work on adding layers of checks, managing outliers, and monitoring the whole process daily. This ensures that our output is the best we can for the given data, but it’s not an end in itself. Customer feedback and end-result testing allow us to elevate our Data quality standards day after day.
Meet
LEARN MORE
[CUSTOMER EXPERIENCE] Discover how IndiGo Airlines seamlessly integrated diverse data sources, such as ACARS, flight planning, FDR, load interim, and AOC data, into the SkyBreathe® platform and the rigorous validation process that ensured data accuracy and reliability. >>> Watch the full interview!
