Design Thinking Approach for Dimensionality Reduction using Ensemble Gradient Boosting Algorithm

• Published in: 2023 5th International Conference on Inventive Research in Computing Applications (ICIRCA) pp. 1468 - 1475
• Main Authors: Kumar, N. Selva, Kalaivani, D.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 03-08-2023
• Subjects: Boosting; Data analysis; Data Analytics; Data models; Design Thinking; Dimensionality reduction; Feature Engineering; Human activity recognition; Internet of Things; Machine Learning; Machine learning algorithms; Medical services
• DOI: 10.1109/ICIRCA57980.2023.10220873

Human activity recognition (HAR) has developed an essential application in the Internet of Things (IoT) domain, providing a foundation for many healthcare, fitness, and security applications. However, IoT sensors' high volume of data generation can make HAR computationally challenging. Dimensionality reduction techniques have been proposed to address this challenge, however, their performance is dependent on a number of criteria, like data quality, algorithm selection, and feature space design. This paper propose a design thinking approach for dimensionality reduction in enhancing HAR in IoT using data analytics and machine learning algorithms. This research work present a comprehensive framework that incorporates user-centric design, data exploration, feature engineering, and machine learning modeling to develop the accuracy and efficiency of HAR systems. The performance of this method is assessed using a publicly available dataset for HAR and results show that our framework achieves a significant reduction in feature space with minimal loss of information, leading to improved accuracy value and efficiency of HAR systems. This research explores into the application of design thinking in addressing dimensionality reduction difficulties in HAR, which can have important implications for the development of efficient and user-friendly IoT applications. The proposed approach aims to create more efficient and user-friendly IoT applications in areas such as healthcare, fitness, and security.