Introduction: Customers purchase items in a supermarket (see Fig. 1a). This purchase data represents big sales data (see Fig. 1b). Useful information that can facilitate the supermarket owners to gain competitive advantage lies hidden in the sales data. Our research investigates novel pattern mining techniques (see Fig. 1c) to uncover customers' purchase patterns (see Fig. 1d) in sales data. The generated information can be beneficial to the managers for various purposes, such as product placements, inventory management, and introduction of new schemes.

Research: Our research aims to discover various types of patterns hidden in sales data. Some of the interesting patterns that we try to discover in sales data are as follows:
(i) Frequent patterns: itemsets that were frequently purchased by the customers
(ii) Rare patterns: Expensive items that were infrequently purchased by the customers
(iii) Periodic-Frequent patterns: itemsets that were regularly purchased by the customers
(iv) )High utility patterns: itemsets that have generated high revenue for the manager

Real-world application: We have applied our models on the Yahoo! Japan retail data and discovered useful information.
(i) We have discovered that many customers were simultaneously purchasing games of Nintendo3Ds and PlayStation. This information is interesting because it disproves the general assumption that people who purchase Nintendo3Ds games may not purchase Playstation games at the same time.
(ii) Many people were purchasing the items, WhitewhineSet and RedWhineSet, in the morning.

Publications and Patents:
(i) R. Uday Kiran, T. Yashwanth Reddy, Philippe Fournier-Viger, Masashi Toyoda, P. Krishna Reddy, Masaru Kitsuregawa: Efficiently Finding High Utility-Frequent Itemsets Using Cutoff and Suffix Utility. PAKDD (2) 2019: 191-203
(ii) R. Uday Kiran, Pradeep Pallikila, José María Luna, Philippe Fournier-Viger, Masashi Toyoda, P. Krishna Reddy: Discovering Relative High Utility Itemsets in Very Large Transactional Databases Using Null-Invariant Measure. IEEE BigData 2021: 252-262

Introduction: Disasters, such as earthquakes and typhoons, are very common in Japan. Recommending proper traffic routes by identifying future congestion patterns is a crucial task to save human life and enhance drivers' experience in autonomous driving. Our research (see Figure 1) primarily aims to develop an explainable AI model to predict future traffic congestion patterns in transportation networks.

Research: Our research aims to discover different types of interesting patterns found in congestion data:
(i) Development of explainable traffic congestion prediction system.
(ii) Discovering the neighboring road segements in which people have regularly faced traffic congestion.
(iii) Finding high congested road segements.
(iv) Designing a data warehouse to store large scale traffic congestion data.

Publications and Patents:
(i) R. Uday Kiran, Sourabh Shrivastava, Philippe Fournier-Viger, Koji Zettsu, Masashi Toyoda, Masaru Kitsuregawa: Discovering Frequent Spatial Patterns in Very Large Spatiotemporal Databases. SIGSPATIAL/GIS 2020: 445-448
(ii) R. Uday Kiran, Sadanori Ito, Minh-Son Dao, Koji Zettsu, Cheng-Wei Wu, Yutaka Watanobe, Incheon Paik, Truong Cong Thang: Distributed Mining of Spatial High Utility Itemsets in Very Large Spatiotemporal Databases using Spark In-Memory Computing Architecture. IEEE BigData 2020: 4724-4733
(iii) R. Uday Kiran, Palla Likhitha, Minh-Son Dao, Koji Zettsu, Ji Zhang: Discovering Periodic-Frequent Patterns in Uncertain Temporal Databases. ICONIP (5) 2021: 710-718

Note: This is a joint collaborative work with NICT X-Data platform.

Introduction: Air pollution is a major cause of cardio-respiratory problems for people living in Japan. To tackle this problem, the Japanese Ministry of Environment has set up the Atmospheric Environmental Regional Observation System (AEROS) constituting of several air pollution measuring sensors (or stations) positioned throughout the Japan. The spatial locations of these sensors is shown in Fig. 1a. The data generated by this sensor network is shown in Fig. 1b. Useful information that can facilitate the ecologists to come up with location-specific pollution control policies lies hidden in this data. Our research aims to transform this big data into a spatiotemporal database (see Fig. 1c) using ETL techniques, apply pattern mining techniques (see Fig. 1c), discover the high polluted areas (see Fig. 1d), and visualize them (see Fig. 1e).

Publications and Patents:
(i) Pamalla Veena, Penugonda Ravikumar, Kundai Kwangwari, R. Uday Kiran, Kazuo Goda, Yutaka Watanobe, Koji Zettsu: Discovering Fuzzy Geo-referenced Periodic-Frequent Patterns in Geo-referenced Time Series Databases. FUZZ-IEEE 2022: 1-8
(ii) Pamalla Veena, Sai Chithra Bommisetty, R. Uday Kiran, Sonali Agarwal, Koji Zettsu: Discovering Fuzzy Frequent Spatial Patterns in Large Quantitative Spatiotemporal databases. FUZZ-IEEE 2021: 1-8
(iii) R. Uday Kiran, Palla Likhitha, Minh-Son Dao, Koji Zettsu, Ji Zhang: Discovering Periodic-Frequent Patterns in Uncertain Temporal Databases. ICONIP (5) 2021: 710-718