Machine learning has taken a significant role in many data initiatives today. Facebook, for instance, is using machine learning to offer personalized ads, whilst Google uses it to learn about its users, and other technology companies are now able to crunch data in a fraction of the time.
Organizations have been looking at machine learning as something that has the most use in looking at optimizing its current markets, but this may not be the case for too much longer. Several companies are now using machine learning combined with predictive analytics to help expand into new markets and exploit opportunities as soon as they come up.
We heard about this from Wolf Rendall, Data Scientist at Auction.com, at last year’s Social Media & Web Analytics Innovation Summit. He discussed how machine learning is now taking a leading role in helping to predict locations, times of the year, and other aspects that will affect the housing market and hence have an impact on the site. He was finding that some areas of the site were drying up, while others were being flooded by houses; and whilst some houses were very popular, it meant that others were being ignored.
To fix this, the team at Auction.com looked at trying to create a recommendation engine, similar to that used by Amazon. However, rather than looking at what might compliment what had previously been bought, it was looking at substitutes for what may have been missed. By looking at significant amounts of user data, it was possible to create personalized recommendations based on a number of factors, creating a far more robust and useable system.
Looking at a broader dataset allows companies to make powerful predictions, and when additional data sources are factored in, they become incredibly powerful.
A prime example of how powerful machine learning can be leveraged is in Utilities industry.
By studying the use patterns of specific areas or even individual customers, it is possible to predict the services they could most benefit from to both save them money and improve environmental issues. Drawing in information from the growing number of smart meters could see utility companies creating one terabyte of new data per day. This could even be mixed with other forms of data, such as heat readings from particular areas indicating poor insulation, or faulty electrical appliances. Companies like BuildingIQ, WegoWise, and Ecova have helped companies save millions of dollars by using this kind of data, and with more available, the market for this kind of work is only going to grow.
Another potential area where this can have a big impact is in Healthcare, where machine learning can help to identify where new healthcare facilities should be placed or where specific units could operate most effectively. For instance, Additive Analytics have been working on a machine learning model to help predict hospital readmissions and trying to predict emergency room admissions before they happen. If this could be used across a population, detailing where people come from for specific treatments, it would be possible to strategically place healthcare facilities where they are most needed, giving better patient care and creating better ROI for invested companies.
Similarly, using machine learning and predictive modeling to identify when people with specific conditions are most commonly admitted may allow hospitals to be best prepared for any potential influx of patients. By doing this, healthcare providers drive better care for patients whilst also cutting down on the potential costs of hiring in temporary staff or bringing in full-time staff when they would be under-utilized.