Data mining to predict the future of business

Associate Professor Hisashi Kashima
(Department of Mathematical Informatics)

Associate Professor Kashima studies “machine learning,” a branch of artificial intelligence, focusing on “predictive modeling,” which is important for business. Application of predictive modeling is targeted at data mining that enables discovery of useful knowledge from data. There exists “analytical” modeling in data analysis to understand what is happening in data but Associate Professor Kashima focuses on “predictive” modeling, which predicts the future based on data.

Prediction result of whether customers purchase a particular product or not has direct impact on a company's sales and 1% of improved prediction accuracy leads to significant profit increase and cost reduction. It is also critical to predict characteristics of chemical substances for new drug development, for example. His research pursues the development of a methodology to predict the future based on a network structure connecting dots and lines, as in chemical formulae that express the form of chemical substances. If successful, applicability of data mining will greatly increase.

Given that data collection and storage infrastructures are in place today, the interest among many companies has shifted to adding values to their products and services by capitalizing on collected data. Companies have started to understand that data analysis technology is a key for business differentiation to enhance competitiveness of their products and services. On the other hand, it is necessary to foster specialized knowledge and capabilities to leverage data analysis technology in the right way.

Associate Professor Kashima aims at bridging business and advanced research based on 10 years of research career and know-how in industry. His research embraces perspectives of both industry and academia.