Data analytics takes flight at Air Canada

At Air Canada, safety is paramount. And for the man leading the data strategy aimed at improving this key performance metric, how insights are delivered into the airline’s safety performance is just as important.

“Where I see myself, and where I invest most of my time, is creating different versions of data for different media,” said Shaul Shalev, Air Canada’s manager of safety analysis and innovation.

[ Lisez la version française: « L’analyse des données prend son envol chez Air Canada » ]

This philosophy focuses on enabling flight crews to access critical safety information with the push of a button or a voice command. This means going beyond the two-dimensional framework of traditional business intelligence tools to include augmented reality applications connected to data dashboards, smart speakers connected to dashboards to provide information in an audio format, and a smartwatch App that displays KPIs like a daily count of airline claims with something as simple as a glance at your wrist. It’s all part of a strategy to provide data in the form that best suits the needs and preferences of a particular employee.

“We can ensure that all our employees process the data they need in their preferred medium,” says Shalev. “This is where I’m going with this whole BI journey.”

A flying leap into data analysis

These innovations – and all potential future incarnations – would not be possible without a sophisticated data platform underlying them.

Air Canada’s BI journey began a few years ago with a mission to modernize its existing Safety Management System (SMS). The airline, which serves more than 50 million passengers each year and was handling summer peaks of up to 1,500 daily flights before COVID-19, has a fair amount of data to manage. Using the old system, employees had to create safety or hazard reports and then submit them to the appropriate managers, a complicated and time-consuming process that often meant contacting three or four people to produce a single report.

“We had hierarchies in databases, self-referencing tables in databases, which becomes a nightmare from a developer perspective,” says Shalev.

So Shalev and his team set about overhauling the airline’s data operations. “The vision was to push relevant, actionable data to our users, who can be anyone from gate agents to our C-level executives,” he says.

Working with its SMS provider, Air Canada has adopted Sisense for data analysis and has been able to simplify its database structures to create nimble, flexible data models that can adapt to changing conditions and applications. The airline is also applying AI to data processing and operations, including security, as a massive amount of data needs to be processed from multiple sources, from aircraft-generated data at a rate of hundreds of gigabytes per fleet per week to human data – generated data such as B. Security-related text reports.

Before Air Canada accelerated its data processing capabilities, it was in a place where many airlines found themselves: they needed people with the very specialized skills needed to compile the data and reports. “Building it and showing it would have taken days if not weeks to put together specific datasets,” he says.

Air Canada’s data transformation has greatly enhanced Air Canada’s ability to quickly slice data, enabling faster risk analysis, trend recognition, timely processing of data and the ability to transfer and view data at the touch of a button.

“Now it’s available 24/7. Just log in, choose your dashboard, then choose the smartwatch app you want to use on your smartphone, choose the Alexa platform or the Echo platform you want to query with real-time questions. It’s that speed and agility that we can offer our users,” says Shalev.

let data fly

Having senior executives on board from the start enabled a smooth migration to an entirely new data platform and a new way of querying and analyzing the airline’s data. But data modeling presented a bigger challenge, says Shalev.

“It takes that database, massages it, creates the new different models, and then connects it to external entities like a dashboard or a smartwatch app,” he tells CIO Canada, adding that another problem is the sheer volume of data that become available. “Those who are not careful will be overwhelmed with it and lose track of things and even expose themselves to risks.”

While some of Air Canada’s data-related proofs of concept have been on hold during the pandemic, the company is still trying to innovate and challenge how far it can push the data. For example, spreadsheets with data aren’t for everyone. Instead, to accommodate people’s different preferences and different abilities to interpret information in a meaningful way, Air Canada is using toolsets in the Sisense platform and developing new applications for presenting and interacting with data, including a mixed reality application the hand gestures control an image of the globe overlaid with relevant information. “On this globe you can see high risk airports and dive into them to see high risk gates and certain KPIs are overlaid on this globe. Bringing in that element of gamification is really going to be big today,” says Shalev.

Providing metadata on the number of aircraft damage incidents and applying algorithms to match incidents to specific airports and specific prevailing weather conditions expands the potential for insights that can be fed back into safety analysis, he says.

Maximizing data analytics applications required a rethink on the part of the airline’s workforce, be it management, gate staff or the developers led by Shalev. “We work hand-in-hand with them to really understand what they want to see,” he says. Shalev likens the process to giving people a blank sheet of paper and asking them to draw what’s most important in their work, and then trying to create it with data insights. “We’re looking at what will result in this blank piece of paper presenting actionable data for those specific lines of business,” he says.

While the focus remains on security, it is guided by their priorities and preferences. As a result, “we are now able to offer people different flavors of the data with much faster iteration.”