A COMPREHENSIVE ANALYSIS AND INSPECTION OF MASSIVE DATA USING DATA MINING TECHNIQUES

Abstract

In addition to storing and retrieving data, modern data management systems sift through massive
datasets to uncover patterns and correlations that were previously unknown. Because new technologies are
developed so quickly, there is an increasing demand for computer applications and data mining tools. The required
tools and software must be able to interact with remote databases in order to guarantee that every calculation yields
the same result. Distributed data mining raises privacy concerns, nevertheless, due to regulatory limitations and the
need for a competitive advantage. This encourages experts in the domains of big data, cyber security, and data
mining to do more research.
Researchers created Privacy-preserving Distributed Data Mining (PPDDM) to address the multi-party computation
problem, in which multiple users attempt to perform a data mining task cooperatively using their respective private
data sets, in order to get around these limitations and benefit from these advantages. Participants discover only the
outcomes of the data mining method and their own inputs after finishing the exercise. The main goal of this study
was to develop a novel way to privacy-preserving data mining for the purpose of developing Decision Tree
Classifiers using vertically partitioned data. Weak is utilized to construct a conclusion tree classifier using the
proposed PPDM algorithm, and the outcomes are contrasted with the well-researched J48 approach. This analysis
employs accuracy and precision. as its standards. Compared to the conventional approach, the suggested PPDM
algorithm offers far greater accuracy and precision. Safe for privacy with the use of modern big data mining tools,
data mining is possible. To determine which Big Data Mining Tool is the best, we examine several possibilities.
The benefits and drawbacks of each instrument are compared to one another. Using Decision Tree Classifier, we
experimentally assess these three methods on two datasets from the online UCI Repository.