Business Intelligence (BI) refers to a comprehensive approach that enables corporations and organizations to perform decision-making based on data utilization. In the Business Intelligence process, various data relevant to decision-making is initially gathered, integrated, and processed. Following this, data analysis is conducted to derive insights by extracting trends and patterns from visually accessible graphs and summary tables. The software and applications that support these processes efficiently are referred to as Business Intelligence tools.
As previously mentioned, Business Intelligence tools support the collection, processing, analysis, and visualization of data. However, here we introduce the Business Intelligence tools specialized in analysis and visualization that our company manages. Our company has developed a Business Intelligence tool based on the "Tensor Self-Organizing Map" technology from Kyushu Institute of Technology. This tool models high-dimensional data in low-dimensional spaces and visually interprets data relationships using heatmaps. Utilizing this tool allows for an intuitive understanding of latent data patterns, which facilitates the exploration of data-driven hypotheses. Moreover, due to interactive operations on the heatmap display and the ability to perform automatic clustering with any chosen square number, it is possible to instantly examine data relationships and compare differences between clusters.

The consulting approach for this tool can be broadly classified into three steps: 'Model Design,' 'Model Construction,' and 'Model Results Confirmation.'

Firstly, in the Model Design phase, we contemplate which data you possess should be used to construct your unique model, what kind of data should be clustered, and consider the final display screen where the model will be presented. Next, during the Model Construction phase, the deliberated content is configured on the BI tool, and computational processing is executed to model the high-dimensional data in low-dimensional spaces. Lastly, in the Model Results Confirmation phase, we visually verify the outcomes of the constructed model through the heatmap display and conduct exploration for hypotheses induced by the data.

Following this approach, once you have grasped the operation and configuration methods of this tool, it is possible to continue using the tool independently.

• By simply clicking on a single drawing result, you can instantly capture the relationships between multiple sets of data on a heatmap.
• Enables speedy in-depth analysis and the exploration of hypotheses derived from data.
• The tool can automatically generate a number of clusters based on any squared number of your choosing.
• It is possible to make predictions for overall data based on the data constituting the clusters.

• Realizing a more humane society through the functional extension of fragrances

• Joint research with IM Corporation, provider of one of the largest public DMPs in Japan