School of Engineering

Graduates the engineering leaders of tomorrow...

Software Design and Implementation of Mirrored Dendrograms Data Visualization Tool

This project describes the software design and implementation of Mirrored Dendrogram, a new unsupervised feature-based tool for data visualization. The tool accepts as input semi-structured data, and allows the user to select the target features to be visualized and mapped against each other, as well as their relative weights reflecting their impact on the visualization process. After inputting the required features, appropriate similarity computations are applied on the data samples based on the user-chosen features and the data types. Then, a hierarchical clustering process is invoked to cluster these data samples and produce a dendrogram structure for each combination of target features. The system decides on the best zooming level to display the dendrograms, providing a compromise between a meaningful visualization high-living the similarities among the data, and an acceptable level of details or granularity. The dendrograms are then mirrored against each other, where the nodes of one dendrogram are mapped against the other dendrogram’s nodes following am adaptation of the transportation optimization problem, in order to highlight their structure correlation. We design a Model-View-Controller architecture to satisfy the functional and non-functional requirements of the project. Moreover, our implementation caters for non-expert users by providing a simple Graphical User Interface (GUI) developed using Python3 and its Tkinter library. The software is not limited to processing EHRs, but it can be used to process with any type of semi-structured data.   Report in pdf

2021-MirroredDendog-JT-res2.jpg


Copyright 1997–2021 Lebanese American University, Lebanon.
Contact LAU | Feedback