Algorithm Development Of Bidirectional Agglomerative Hierarchical Clustering Using Avl Tree With Visualization

• Main Author: Hussain Mohammad Yousef Abu Dalbouh
• Format: Thesis
• Language: English
• Published: Universiti Sains Islam Malaysia 2015
• Subjects: FST
• Online Access: http://ddms.usim.edu.my/handle/123456789/8226

In recent years, the dramatic rise in the use of the internet and the improvement in technology in general have transformed societies into one that strongly depends on information and knowledge. The growth of information resources along with the accelerating rate of technological change has produced massive amount of data and information that often exceed the ability to handle and manage it. Therefore, the demand now is creating a faster approach to handle voluminous data. This will also improve the complexity time of the traditional hierarchical methods to face huge collections of data and growing information flooding. In addition, user involvement in the data mining is needed as whereby the user interact with the process through exploitation of the power of human explanation sight and brain for analyzing and exploring data. Clustering is an analysis technique for discovering interesting distributions and patterns in the data set. The objects within a cluster are more similar to each other than the objects in different clusters. This research proposed a bidirectional agglomerative hierarchical clustering algorithm. The proposed algorithm is fundamentally similar to conventional agglomerative hierarchical clustering algorithms designed to partition a collection of objects into subsets sharing similar attributes. It is obvious that analyzing large data sets via traditional methods has moved from being tedious to being high computational cost. The traditional methods usually not scalable to very large datasets, with an O(n²) computational cost. However, the proposed algorithm adapted AVL tree approach cluster the objects to left and of right the median/root. The computational cost significantly reduced into O(log n). This is efficient for huge amount of data. Thus clustering using bidirectional hierarchical will facilitate efficient computational cost. This research demonstrated the agglomerative algorithm performance based on complexity parameters such as execution time and the number of cluster needed to merge all data point/objects into one cluster. As part of the experimental validation, real data set were used to measure the effectiveness and the efficiency of the proposed algorithm. The study shows a 73.4% improvement from the traditional approach. The demand for visual and interactive analysis tools is particularly pressing in this information age, where the user needs to analyze and observe large amount of data to grasp valuable knowledge. This research also proposed a visual cluster approach to visualize the knowledge extracted by the data mining algorithm using AVL tree approach. The visualization prototype is evaluated by postgraduate students who were interviewed and using Technology Acceptance Model, as the instrument. The result revealed that visualization is useful, easy to use and give user satisfaction.