Drowning in data

I went out for my usual run this weekend, wearing my GPS-enabled watch with a heart rate monitor. I use the information provided to give me an overall view of how I am progressing, if at all, or even regressing, which is more frequently the case these days. Unfortunately, the device does not slow the ageing process. I only look at the data when I return; I try to enjoy the run free of any data updates. When I get back, I have a plethora of data to peruse and compare: the time of day of the exercise; overall distance; overall pace; calories burned; average heart rate; maximum heart rate; training effort; and cadence. Then, for the pre-set lap distance, I can examine the same data lap by lap.

A few months ago, following a chest infection and on the doctor’s recommendation, I bought a pulse oximeter that fits on the finger and reads the pulse rate and the level of oxygen in your blood. There are similar wrist-worn devices that also analyse sleep to predict day-to-day health and exercise performance. If I undertook this analysis before exercising (which I do not) and then did a deep dive into all of my post-exercise data (which I do not), I would probably spend three times longer on data analysis than on exercising. This could be increased further if I used an app that uploaded my data and allowed comparison with that of other people.

Sport has obsessively adopted the collection and analysis of data. You can get a tennis racquet that monitors your shots, providing information on the number of forehands and backhands and how hard the ball is hit. It also compares the path of the swing to the ideal. I am writing this with the TV on in the background showing the Six Nations rugby. The aficionados among you will know that rugby teams often spend time swapping kicks to gain territorial advantage. In an attempt to make this more interesting to the viewer, the territory gained by the kicks is now displayed courtesy of sensors in the rugby ball. These also monitor the speed, the hang time and the spin speed. All of this data may be of benefit to elite athletes and coaches, but is of more dubious value to mere mortals and can detract from the enjoyment.

Not all data is useful, especially when statistical analysis is applied. While waiting for a recent kick on goal, we were informed that the kicker had a 71% success rate from this distance and the average for all kickers was 67%. So what? In this case, previous performance did not determine what ensued. Despite being better than average, the kicker missed. To derive benefit from statistical analysis, it is important to understand its underlying limitations.

During the Vietnam War, American Defense Secretary Robert McNamara was driven by numbers and the logic of mathematics. He tried to model the war with factors that could be measured, including the infamous body count. Edward Lansdale told him that it was impossible to measure the most important variable: the human factor, which was too hard to account for, and the rest is history. Accounting for human factors is still a challenge for non-destructive testing (NDT). We can establish the physical capability of various NDT techniques and have used this in establishing measures of the probability of detection, but there is still some way to go to be able to account for the human influence on this metric. Work is ongoing.

But what of other NDT data? How well are we doing? Technology now allows immense amounts of data to be collected and displayed. Are we keeping up with interpretation and analysis? The limitations of early technology forced designers to identify information that was important for successful application and to design collection and analysis processes accordingly. Now there appear to be no such drivers and the default condition is to collect all possible data and worry about turning it into useful information afterwards. Machine learning and artificial intelligence are tools that can help manage this situation, but it is important that the underlying limitations are fully understood.

The challenge for the NDT profession is to do the upfront thinking, so that all data collected provides useful information and is not collected just because it can be. This will help us to avoid redundant outputs such as the internet-enabled washer dryer that records the user’s favourite wash cycle and the percentage of occasions it is used!

Please note that the views expressed in this column are the author’s own personal ramblings for the purpose of encouraging discussion within NDT News. They do not represent the views of Jacobs or BINDT.

Letters can be mailed to The Editor, NDT News, Midsummer House, Riverside Way, Bedford Road, Northampton NN1 5NX, UK. Email: ndtnews@bindt.org or email Bernard McGrath direct at bernard.mcgrath1@jacobs.com

Comments by members

This forum post has no comments, be the first to leave a comment.

<< Back