Researchers have developed a model that predicts the chances of success and failure of a business venture within six months with 80% accuracy. The information that led to the formation of this model was taken from the social media and transport data over the past few years.
The business environment of today’s world is very dynamic. Every day thousands of businesses either shut down or open up. The retail sector is obviously a very dynamic and risky sector too. This is mainly because over the past few years the shopping patterns of consumers have changed dramatically owing to the new trend of online shopping and a change in the way people spend. Obviously, these are not the only factors affecting the retail stores.
PhD students at Cambridge University and Singapore Management university wanted to account for and understand these factors better. In order to do that, they built a machine learning model which could predict with 80% accuracy, the shop closures in ten cities around the world. The model studied how people move in urban areas and how the location of a business affects its success. This research was obviously to help business owners in taking the best decision regarding their business’s location, opening hours etc.
They used data about consumer’s demand and transport and then devised a metrics in accordance, with the help of which, the machine learning model identified patterns. After that it was analyzed how well the model could predict the success of a business once it was given metrics of the area the business was in.
The data came from Foursquare which is a location recommending app, information about anonymous users 74 million check-in details was gathered which represented the demand of consumers. Also data from 181 million taxi trips, their pick off and drop off points in New York and Singapore to identify how people moved across the city.
Although there are many factors affecting the success or demise of a business, According to Krittika D’silva, the Ph.D. student at Cambridge University, using Venue specific, location related and mobility based features to predict if the business could succeed or simply fail.
It was found that different metrics affected different cities. But for Chicago, London, New York, Singapore, Helsinki, Jakarta, Los Angeles, Paris, San Francisco and Tokyo it was found that the range of time matters a lot. Business who only do well in a particular time slot doesn’t survive long. While businesses who perform other than the typical hours tend to survive longer. It was also discovered that when the diversity of a business declines, the chances of survival tend to be very low. The results also suggested that businesses that exist in a diverse neighborhood also tend to survive longer.