Researchers from the Graduate School of Information Science and Technology at the University of Tokyo created an algorithm that can predict consumer purchases. Brands could use the algorithm to analyze their potential customers and create marketing strategies.
A representation of a statistical network researchers used in their algorithm. Source: 2020 Yamasaki et al.
To create the algorithm, the team combined statistical modeling technology with machine learning-based image recognition. They used various computational tools to data-mine social media.
In the past, companies used customer surveys and sales projections to improve their marketing strategies. This is time-consuming and imprecise, leading to unsuccessful marketing.
The team started their work by gathering publicly available social media data from followers of brands. They used image recognition and machine learning to analyze and categorize photos and hashtags that relate to a brand’s followers. This revealed consumer behavior patterns in relation to brands, which allowed researchers to calculate the similarity between brands.
The algorithm was evaluated against purchase history and questionnaires. Questionnaires provided context on the customer’s purchase information. Experimental results showed that credit card and point card companies could predict customers’ past purchasing behavior. The algorithm accurately predicted customer willingness to try new brands. It proved useful for brands to create new promotions with social media. It could also be useful for shopping centers and malls to plan which stores to include and what brands to stock. Brands could also use the algorithm to match their brand with social media influencers for collaborations.
A paper on the algorithm was published in the Proceedings of the ACM Multimedia Asia.
