The algorithm learns from the data instead of using any model assumptions
The best place to live in the world is West Perth, Australia, according to a machine learning algorithm used to identify and rank the best seven places globally.
The findings emerged from the Paradise Found project that analysed nearly 150,000 locations worldwide in 193 countries, comprising more than five million data points.
Software giant SAS said it used its capabilities in artificial intelligence to create the list from large volumes of publicly available data, including city studies, social media sites, review sites like TripAdvisor, geodata and reports from statistical services and international agencies.
Eight different categories were generated by machine learning including: living expenses; safety and infrastructure; healthcare; restaurants and shopping; the environment; culture; attractiveness to families; and education and employment.
A machine learning process used all the publicly available data analysts were able to obtain and an algorithm then determined its importance. The algorithm learns from the data instead of using any model assumptions, so is the sole arbiter of the factors that describe a location – and then uses this information to determine which is the best, said SAS.
The UK market town of Hebden Bridge in West Yorkshire also featured in the best seven places globally, ranked at number five. The full list is as follows:
SAS said when putting together a conventional survey, it is all too easy for unconscious bias to creep in when selecting the criteria to use when determining which data should be collected and analysed. For Paradise Found, it processed all the available data and allowed machine learning algorithms to decide which criteria were important.
“The data doesn’t lie,” said John Spooner, head of data science at SAS UK & Ireland. "[In] this way, no aspect can be ignored simply because no one was looking for it.
“[The methodology] allowed us to demonstrate what analytics and machine learning are capable of -- namely, finding patterns in data from a completely impartial perspective. In this particular case it showed how analytics can come up with a list of places that are different to what people might first think of based on their own opinions and preferences.”
If you like this, you might be interested in reading the following:
Citizen engagement is key to smart city success
By 2020, two-thirds of all smart city execution strategies will incorporate key performance indicators
Partnership aims to build more resilient communities
The organisations will jointly develop a number of smart city knowledge resources and showcase leading practice solutions
SAS pioneers smart campus tech
The analytics leader aims to lead the way through lower energy consumption and greater operational efficiencies