[2P4-GS-10] AI application: marketing / optimization

[2P4-GS-10-01] Use purchasing data and movement data to build a model that proposes a user with a high possibility of interaction.

〇Junya Suzuki¹, Kazuki Hirahara³, Naoki Igarashi⁵, Mikiya Morisaki⁴, Hirofumi Nakayama² (1. kojin, 2. Tokyo University of Science, 3. JGC Corporation, 4. Department of Information Science and Engineering, Faculty of Engineering, Yamaguchi University, 5. Deep Percept Inc.)

[[Online]]

Keywords:AI, recommendation, Purchasing data

Background / Issues: Due to the impact of the Corona disaster, people's exchanges are decreasing, leading to a loss of opportunities for innovation creation. Participation in opportunities to have connections with the outside world has high hurdles, and exchanges do not proceed as expected. It is necessary to expand the business exchange environment between different industries and actively discover seeds for collaboration between companies.
Purpose: To support exchanges by matching customers who have a high possibility of interaction from the behavior patterns of purchasing and movement. It creates an opportunity to build a "bridge" that transcends companies and to create innovation.
Outline of measures: Predict user behavior using customer purchasing data and movement data. Similar members are recommended using the predicted behavior pattern as a feature.
result:
- We created a movement forecast AI and a demand forecast AI for each building as models for creating features, and verified the accuracy according to the specified evaluation criteria.
- Regarding the output result of the final recommendation system, it was confirmed that people with similar purchasing tendencies were correctly recommended in the past data.
- We created a demo environment for the entire system, including the user interface, and confirmed the operation assuming actual use.