A data strategy roadmap to improve emergency response services

by | Mar 25

8 min read

You can have data without information, but you cannot have information without data” – Daniel Keys Moran. In order to live out our vision of creating a world where everyone is safe, AURA needs to ensure the quickest response to an emergency, and this entails adopting and relying on a data-focused approach.

 

Imagine the following scenario: a shop owner in the early 1950s has been granted the contract to supply water at the finish line of a neighbouring town’s international marathon. The town is 50 kilometres away from his shop, and having done the journey several times, he knows it takes him about 30 minutes to get to the neighbouring town’s main square. 

 

He wants to determine how long after the race has started should he leave his shop in order to be there just in time for the first runners crossing the line considering he needs 30 minutes to set up upon arrival.

 

With no library subscription to gather information from historic records, he decides to go to the local sportsground in order to gather some data from the town’s runners. Only two of the town’s runners had ever run a marathon. One took three hours, and the other four. Taking the average, the shop owner assumed the majority of the runners would take three and a half  hours. To cater for the top achievers, he would arrive at the finish line 2 and a half hours after the race had started and have more than enough time to set up in preparation to distribute water to the runners.

 

Little did he know that most professional marathon runners, even in the 50s, completed a marathon in under three hours.

 

The above clearly shows how a small sample set can skew results.

 

When it comes to making a decision, how much data is enough data? Is no data better than little data? From my experience, it depends on the decision to be made, the knowledge of the contextual environment, the business and project resource and time constraints, and the criticality of the outcome.

 

On one hand, if the shop keeper needed to determine the volume of water to provide each runner, he could easily leverage off of his own personal experience and a very small sample set in order to make an assumption. Assumptions and projections can be gold when having limited datasets as long as they are educated and leveraged off of environmental knowledge.

 

If, however, the shop keeper had to determine his cost of supplying the water versus his sales revenue, his estimate in the water volume would need to be a lot more accurate than his assumptions and information above. He would have needed further data to make an informed decision on his quotation when he applied for the contract and taken into account all other data costs such as people, capital and material resources. The water volume estimation is a lot more critical in his perspective when it comes to his own profit margin compared to the estimation when just considering the amount of water required for each runner post a marathon – It would surely differ if runners were asked after completing a marathon. Perspective evidently also has a role to play on the criticality of the decision. 

 

One must also be wary of the shortfall of assumptions and projections on a dataset.

 

Let’s assume the shop owner got five data points from five 1000 metre runners when estimating the time to complete a marathon. The average across all 5 runners comes out to two minutes 30 seconds to complete a 1000 metre race. The shop owner simply assumes the marathon runners would maintain the same pace as the 1000 metre runners across the entire range. The proportional projection estimates that the average marathon runner will complete the race in one hour and 45 minutes. Assuming the best runners complete the race in an hour and a half, the shop owner estimates he would need to arrive at the finish line with his water stock 1 hour after the race begins. This is a whole 30 minutes too early that he could have used for productivity at his shop or to enjoy some family bonding time.

 

However, as mentioned previously, assumptions and projections can also be valuable. If the shop owner took a dataset of 5000 metre runners – where the pace would be a bit more similar to that of a marathon runner across the board – the average would have been 15 minutes for a 5000 metre runner to complete a race. Projecting this average to the length of a marathon proportionally would have given him an estimation of 2 hours and 6 minutes, a lot closer to reality with the current world record being 2 hours 1 minute and 39 seconds. 

 

The effects of outliers in datasets

 

Earlier, we mentioned the shop owner has done the journey from his shop to the town square on a multitude of occasions with it taking on average 30 minutes. His quickest time has been 25 minutes while his slowest has been 35 minutes. Despite the norm, on the day of the marathon, it takes him an hour and leads to a disastrous hurry to offload the water as the runners cross the line. The shop owner did not consider the traffic and road closures caused by the event.

 

While the shop owner could have potentially predicted the traffic and road closure, outliers are typically the occurrence of what seems to be disorderly and irregular environmental events. When it comes to emergency dispatch response times, this could range from the randomness of the weather, the unpredictability of traffic lights to unexpected road closures.

 

Phew… now after that waffle, how does this relate to achieving our ultimate goal of improving emergency response times to protect assets and people?

 

In essence, AURA needs more data in order to improve the smart technology rolled out by our developers, firstly through machine learning with an ultimate goal to reach a state of artificial intelligence in the safe tech ecosystem. 

 

Machine learning enables machines – meaning computers, whether physical or virtual – to learn from data with a specific goal in mind, while artificial intelligence is a broader concept which creates “intelligent” machines in order to try to simulate cognitive human thinking to enable decision making.

 

Even the parameters that seem random and disorderly can be simplified into a chaotic complex system by analysing patterns through chaos theory if enough data is provided and incorporated into machine learning and artificial intelligence.

 

In order to increase the data set and improve safety across the globe, AURA will begin embarking on a data collaboration journey with its ecosystem’s stakeholders while ensuring compliance with the POPI act: 

  • Security and medical service providers will begin sharing information regarding their incidents and resources,
  • Customers will begin sharing its user base demographic details and site information,
  • State police and other safety and crime prevention organisations will ideally also begin sharing their datasets.

 

By amalgamating all these datasets, AURA can begin matching up the supply network of our service providers to the demand needs through machine learning, whilst beginning to train an artificial neural network to make the correct decisions in the long term. The ultimate idea is for the tech to tell our service providers to place their vehicles in specific areas not only to achieve better response times but at the end of the day prevent crime and save lives.

 

If the marathon analogy didn’t quite do the trick, this entire article can be summarised as follows:

 

Imagine buying a 2000 piece puzzle in a charity shop – probably not a good idea in the first place – spending hours trying to complete it, only to find out at the very end that20 pieces are missing. If you knew there were pieces missing, would you have pursued it in the first place? Most people wouldn’t, but entrepreneurs and disruptors don’t allow these blockers to hold them back. They will do their utmost to find the last pieces of the puzzle to unveil the entire picture. Whether the pieces require searching far and wide for, or whether the pieces require creating, the AURA team is dedicated in gathering all the pieces of the puzzle to ensure it achieves its vision: to create a world where everyone is safe.