ROCK ART
Andrea Jalandoni, of Griffith University in Queensland, Australia, is making use of machine learning to analyse vast amounts of data, such as photos and tracings, collected at sites of millennia-old rock art in the Pacific, Southeast Asia and Australia. ‘Manual identification takes too much time, money and specialist knowledge’, Jalandoni says. She worked with Nayyar Zaidi, of Deakin University in Victoria, Australia to test machine learning to automate image detection using hundreds of photos from the Kakadu National Park in Australia’s Northern Territory, some of which showed painted rock art images and others with bare rock surfaces. The system found the art with an accuracy of 89%. Initial results were published last August – see https://bit.ly/RockArt2022.
The rock art is created from pigments made of iron-stained clays and iron-rich ores that were mixed with water and applied using tools made of human hair, reeds, feathers and chewed sticks. Some of the paintings in this region date back 20,000 years, making them among the oldest art in recorded history, only a fraction of the works in the park have been studied. ‘For First Nations people, rock art is an essential aspect of contemporary indigenous cultures that connects them directly to ancestors and ancestral beings, cultural stories and landscapes’, Jalandoni says. ‘Rock art is not just data, it is part of Indigenous heritage and contributes to Indigenous wellbeing’.
10000 STEPS, EXACTLY
How would you like to create a route of a specific length? Rhyd Lewis and Padraig Corcoran of Cardiff University have developed a journey-mapping algorithm to allow you to do just that. Turned into scientific terms, their paper, published in the Journal of Heuristics, is entitled ‘Finding fixed-length circuits and cycles in undirected edge-weighted graphs: an application with street networks’. You can read it at https://bit.ly/RhysandCorcoran2022 or maybe go for a walk.
SUCCESS AND FAILURE
In a first in a series of ten articles, Douglas A. Gray in INFORMS’ Analytics magazine (https://bit.ly/Gray2023) discusses successful Data Science projects and why they fail. He points out that ‘High-tech companies, such as Google, defined as “analytical competitors,” use data science aggressively throughout their entire enterprise to sharpen operational performance and efficiency and improve customer experience in their retail and online search businesses, respectively. Companies like American Airlines pioneered the use of data and analytics in the field of revenue (yield) management in the 1980s to generate $400-$500 million in incremental revenue annually. UPS saves $300-$400 million annually with its On-Road Integrated Optimization and Navigation (ORION) application that guides their 55,000 delivery truck drivers every day. Walmart generates millions of dollars in value annually by applying predictive and prescriptive analytics to optimize its markdown pricing strategy’.
But, despite these, and many others, success stories, ‘according to a study by Deloitte Analytics and Tom Davenport, only 20% of data science models built are actually deployed into a production system supporting a business process’. Why is this? Gray argues that ‘the problem is not with the mathematics and technology but rather with the actions of the people (practitioners and leaders) … and the processes employed to execute and manage data science projects’.
Gray’s remedy is twofold. First, begin with the end in mind: ‘You don’t want to just build a model; rather, you want to embed that model into a mission-critical system that supports a key business process such that greater economic efficiency (i.e. lower cost, greater revenue, improved customer experience) can be achieved on an ongoing basis in an automated manner with little or no human intervention, creating a flywheel effect generating business value’. Secondly, sharpen the saw. By this he refers to appropriate education, so that ‘students, practitioners, leaders and executives “sharpen the saw” and fill in the knowledge gap in their training and education that heretofore was learned only through real-world work experience’.
PREDICTING ASSET HEALTH
An analytics engine created by Viking Analytics, using AI-based algorithms, that automatically detects unseen or pre-failure operational conditions for electrical equipment, makes it easier for operators to prevent costly failures, plan maintenance efficiently and maximise uptime. It will allow customers to predict anomalies before they become a risk to their operations. Rajet Krishnan, CEO of Viking Analytics commented: ‘Our company’s mission is to allow industrial specialists to easily extract insights and value from both process data and asset data through AI’. A strategic partnership with ABB allows Viking Analytics’ technology to complement the ABB Ability Asset Manager, and provide full remote visibility of asset and electrical-system health status.
Sherif El-Meshad, of ABB: ‘With the pressure on to ensure uptime and prolong the lifecycle of electrical assets, the partnership with Viking Analytics allows us to develop analytics that will help customers maintain their operations and cut costs. Customers will get the insights they need to make informed decisions about their electrical equipment fleet and take preventative actions to avoid costly failure’. Read more at https://bit.ly/VikingAnalytics.
CATCHING THE EYE
Inefficient management of resources and waiting lists for high-risk ophthalmology patients can contribute to sight loss. So, researchers at Cardiff University used O.R. to develop a decision support tool to determine an optimal patient schedule for ophthalmology patients. Available booking slots as well as patient-specific factors such as eyecare measure risk factors, referral-to-treatment times and targets, and their locations were taken into account. The model can be applied and implemented without the need for additional software, to generate an optimised patient schedule. Their published paper can be found at https://bit.ly/CardiffEye2023.
THREE MISTAKES AND YOU ARE OUT
An article by Nahla Davies, https://bit.ly/ThreeDataMistakes in Data Science highlights the top three mistakes that companies commonly make that affect the accuracy of their data analytics.
First, Data Cleaning Isn’t at the top of your to-do List. Most sets of data have their fair share of errors. Whether they’re typos, weird naming conventions, or redundancies, errors in data sets muddle the accuracy of data analysis.
Second, the Algorithms you’re using aren’t accurate enough. Most algorithms aren’t one hundred percent perfect; most of them have their fair share of flaws and simply don’t work the way you’d like them to every time you use them.
Third, the Models you’re using aren’t that good. ‘Algorithms can crunch data all day long, but if their output isn’t going through models that are designed to check the subsequent analysis, then you’re not going to have any usable or useful insights’.
HUMANITARIAN AID
A report by the Geneva Centre of Humanitarian Studies (https://bit.ly/ORandHumanitarianAid) argues that there is an increasing and critical need for O.R. in humanitarian settings, because time is invariably of the essence and problems and needs are often ill-defined. Practising O.R. in humanitarian settings is often challenging, but there is need for evidence-based decision-making by humanitarian organisations. O.R. can help in equipping decision-makers, at an individual programme and at the policy level, with the high-quality evidence that is needed. The Centre has gained extensive experience as a community of researchers and practitioners over the years on how to mitigate and adapt to many of the common challenges faced in humanitarian operations.
