A Comparison of Real-Time Data Analytics Algorithms

Authors

ผศ.ดร.เอกรัฐ รัฐกาญจน์

Published

TNI journal of engineering and technology

Abstract

Today’s world is overwhelmed with the stream of data generated by IoT sensors, Smartphone applications, E-commerce transactions, etc. Data streaming and real-time analytics tools are necessary to apply for various purposes such as financial fraud detection, recommended products, or disaster warning systems. Existing real-time data analytics tools such as StreamDM, Scikit-multiflow, or Massive Online Analysis (MOA) play a significant role in this field. There is, however, still a lack of well-comparisons among streaming algorithms in these tools. In this paper, we aim to study and compare the performance of the streaming algorithms provided by Scikit-multiflow, one of the most popular tools. In the experiment, we compare various algorithms on classification and regression problems in terms of accuracy, model size, memory, etc. The synthesized and real-world datasets are both employed for the experiment. The experimental results illustrate that the Hoeffding-Tree algorithm shows the best performance among other algorithms.

(2564). ความตื่นตระหนกทางวัฒนธรรมในห้องเรียนของนักศึกษา ชาวอเมริกันในมหาวิทยาลัยนานาชาติในประเทศไทย. วารสารมหาวิทยาลัยราชภัฏธนบุร, 15(1), 124-130.