Nice data mining!
The relationship between research and practical use and what can be done from an academic standpoint is discussed and the results presented by researchers are difficult for a general person to understand.
Abstract
The statement “Data Scientist: The Sexiest Job of the 21st Century” was printed in the Harvard Business Review about 10 years ago. Since then, data mining and artificial intelligence (AI) have become widespread and have been utilized in research and real world applications. We sometimes get consultations such as, “Please teach me about AI,” “I want to use deep learning in my business,” or “I want data mining to do a good job of analysis. What should researchers do? In response to these consultations, I will discuss three issues: 1) problems in which the data are available but cannot be used immediately (preprocessing issues), 2) problems in which the results are difficult to explain in an easy-to-understand manner (explainable AI), and 3) problems in which the purpose of data analysis has not been identified. First, I consider the topic “when we receive a consultation, there is a lot of data, but it is not ready for use.” This is a case in which the “data ready” stage identified by the researcher is different from the “data ready” stage identified by a general person. The second topic considers the following problem “the results presented by researchers are difficult for a general person to understand.” The researcher's explanations of the algorithm and its results are very difficult. A general person is unable to judge the data analysis result from the viewpoint of “is it useful for business?” The last topic is the most difficult; A general person vaguely thinks that “AI can do something,” but “something” does not allow researchers to identify issues. I will discuss the relationship between research and practical use and what can be done from an academic standpoint.