Building socio-analytical systems to determine user preferences and choices

Imagine if your cell phone provider knew when you were unhappy and offered you a better deal, or if it offered you the right warranties that met your needs exactly.

Ebrahim Bagheri, a professor in the Department of Electrical and Computer Engineering, is looking at ways to help industry predict consumer demand. By analyzing social media data through predictive analytics, professor Bagheri is able to help businesses not only identify their customers but also cater to their needs in a more personal manner.

“Users generate vast amounts of actionable data on social networks, some of which would only be available if time-consuming, expensive field studies and surveys were conducted,” said professor Bagheri. “It would be too impractical to collect a lot of this data with social networks.” Instead, professor Bagheri is working on computational models that gather and analyze the data.

Professor Bagheri's work analyzes the underlying phenomena behind how social data is generated, looking for clues that indicate how individuals are leaning in their decision making. His system also mines for life events such as marriage, the birth of a child, or a change in career, which might help predict purchasing habits. He is further able to analyze what people talk about on Twitter, using that knowledge to predict, for example, if they will react favourably to a new product or to a change in policy.

“The findings can be applied at a microscopic level — examining the individual user. We can determine what product a person might buy based on what they said on the social network after having experienced a certain life event,” he said. “Or we can apply it at a macroscopic level — examining communities and like-minded individuals. We can find people with similar preferences and interests to make generalizations that can lead to interesting and accurate predictions.”

With those predictions, professor Bagheri’s industry partners can prescribe action to get a favourable outcome. For instance, one of his partners is a protection plan provider that is using this information to find out what would be the most desirable features of an extended warranty program. By tailoring the product, the provider can make it more appealing for more people to purchase.

Professor Bagheri’s work is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC), the Canada Research Chairs Program (CRCP), Ontario Centres of Excellence (OCE), and Southern Ontario Smart Computing Innovation Platform (SOSCIP).