- Like Post
6 things you didn’t know data science could do
From fighting crime to uncovering the secret to human happiness, here are six amazing things people are doing with data right now.
One of the things that always astounds me about our work is the seemingly infinite applications for data science and analytics.
At The Smart Cube we work for all sorts of clients in all sorts of industries, solving all sorts of problems. We help identify customer needs, choose the right product mix, enhance supply chain efficiency, and use sentiment analysis to judge product success.
We do a lot. But in the grand scheme of things, it’s only the tip of the iceberg. Outside of The Smart Cube, and data’s numerous business applications, there are countless exciting and surprising things analytics is being used to achieve.
I thought I’d take the time to look at just a few of them.
Finding the secret to happiness
For centuries, people have debated the secret to happiness and come up with all sorts of conclusions, from oneness with nature to flexible working.
It might seem like a fairly nebulous concept to pin down. But wellbeing is intrinsically linked to better health and higher productivity, so governments have a real incentive to identify what makes people happy.
The good news is, data may hold the answer.
With the aim of learning when US citizens were at their happiest, a team of scientists recently used sentiment analysis to examine millions of books published between 1820 and 2009.
The resulting study uncovered several takeaways, including the fact that increases in national income generate increases in national happiness, and that living in peace time has a dramatic impact on wellbeing – the equivalent of a 30% rise in GDP.
Translating lost languages
Google’s AI research arm DeepMind has made headlines several times, by beating the world’s best chess and Go players.
In collaboration with the University of Oxford, a team of computer scientists recently trained a set of neural networks to recognise words on Greek stones up to 2,600 years old. The neural networks were then tasked with predicting missing characters on damaged relics, and were able to do so with nearly 70% accuracy – all in just a few seconds.
To put this into context, human historians only achieved 43% accuracy after two hours of work.
At The Smart Cube, we are working with a number of clients on computer vision projects. One recent use case involved extracting human emotions from video data, based on facial expressions, so it is interesting to read about a new breakthrough from Fujitsu Laboratories which has taken things up a notch.
Computer vision can now more accurately track complex facial expressions like awkward giggles, nervousness or confusion. In contrast, traditional facial recognition tools are mostly limited to eight states of emotion: anger, contempt, fear, disgust, happiness, sadness, surprise, or neutrality.
With this new technology, data scientists can detect emotional changes as subtle as nervous laughter with an 81% accuracy rate. Compare that with Microsoft’s facial recognition technology, which only has a success rate of 60% with more easily detectable expressions.
Transforming film production
Making film and TV is an expensive, time-consuming and complicated process. But deep learning is on the verge of making many elements of that process easier.
For example, Google has developed a neural network that can automatically separate the foreground and background of a video, meaning green-screen technology can now be done by pretty much anyone, without the need for special equipment.
We’ve also seen deepfakes used to edit actors into films they never appeared in, which could have huge implications for the personalisation of films in the future. And perhaps most impressively – or frighteningly – AI may soon replace film producers altogether.
It’s not all about looking forward, though. Machine learning can also be used to restore old films, colourise black and white footage, and even recreate missing frames in footage by analysing the beginning and end of a shot.
Human trafficking is a $150 billion business with more than 40 million victims a year and a chain of exploitation made up of around 172 countries. It’s a huge problem – and if there’s one thing data is good at, it’s solving huge problems.
This year, Stop the Traffik has partnered with the Edelman Predictive Intelligence Centre to build a human trafficking prevention model with insights into market supply and demand, financial information, and trafficking routes.
And this isn’t the only way data is being used to fight crime. Another example is the ‘pre-crime’ initiative led by the Chicago PD, which sounds for all the world like something from Minority Report.
This project uses machine learning, predictive analytics, police databases, and IoT data to pinpoint problem locations – helping the PD to understand the conditions which are most likely to lead to crime.
The advent of industry 4.0 has seen a big rise in predictive maintenance, where analytics is used to foresee the failure of machines well before it takes place.
With the data needed to do this not always readily available, the latest developments have seen maintenance personnel use thermal data and image processing to make predictions about the health of their assets.
According to a recent blog from BCG Gamma thermal imaging “provides a way to perform indirect measurement that in combination with advanced data analytics will provide all the predictive maintenance information needed to avoid unnecessary shutdowns.”
This can be particularly effective in manufacturing environments, where any disruption can incur major costs.
Want to learn more about the varied uses of data science? Read our blog on how analytics is transforming healthcare.
To learn more about what else is new in the world of data science and the projects we’re working on, explore our blog.
And read about how our solutions can help solve your business problems with data and analytics.
Prasad Kothari is an analytics and data science leader who has worked extensively building high-performing teams for various organizations and has provided consulting to many fortune 500 clients. As vice president of analytics and client solutions at The Smart Cube, he focuses on helping clients realize the value of data science to solve priority business problems, including customer analytics, marketing analytics, RWE/RWD, and supply chain analytics. Prasad has published healthcare data science research papers across leading journals, as well as books on AI. He has collaborated with several universities in the US and given guest lectures on Quantum Machine Learning, NLP/NLU, topological data analysis and computer vision research. He spends his weekends reading AI books and listening to Indian classical music.