Database Performance Evaluation and Applications of Data Science for IoT Platform Analysis

Abstract

Big data is very important for businesses and society because for the competitiveness of organizations, properly storing and managing large amounts of data enhances insight, decision-making, and process automation. Application of large storage technology, data mining and analysis is a research direction with practical needs when testing on IoT data sets. Deciding which tool to choose depends on the nature of work. One of the ways to evaluate the system towards data is to conduct the test in environment which the database will run, under the predicted data and the user's current workload. This paper comparatively studies on performance of opensource databases presenting in Internet of Things (IoT) systems such as NoSQL database-based MongoDB and SQL database-based PostgreSQL. Data stored in the database will be exploited, processed, and analyzed. Mining and analyzing data are one of the key parts in the data science. Data science is an interdisciplinary field of study that encompasses processes and systems for extracting knowledge within data in various structured and unstructured forms. Data science includes many areas of data analysis such as statistics, machine learning, data mining, predictive analysis. To provide the necessary rationale for the plans, it is to build a predictive model. Without predictive analytical tools we would be overwhelmed with data without information, not knowing what happened next. Based on the content, method and purpose of the forecast is divided into two categories: qualitative methods and quantitative methods. Qualitative methods often depend heavily on the experience of one or more experts in the relevant field. Quantitative method uses historical data over time, based on historical data to detect the movement direction of the object that is suitable for a certain mathematical model and at the same time use that model as estimation. This research work also presents analyzing and building time series prediction model.