mds-Rose

Overview

mds-rose is an experimental data visualization application where users can upload their datasets and select features to analyze. It uses MDSJ algorithm to reduce dataset dimensions to 2D and visualizes results on a plot.

Features

• Dataset upload
• Features selection
• Euclidean distance calculation
• Outputs 2D projection

Installation

Prerequisites

• Java 23 or later
• Maven (for dependency management)

Build & Run

Clone the repository:

git clone https://github.com/dwdGit/mds-rose.git
cd mds-rose

Compile the project using Maven:

mvn package

Run the application:

java -jar target/mds-rose-<version>.jar

Usage

1. Signup or Login.
2. Dataset upload with .csv or .data extension.
   • The dataset should be structured with numerical feature columns and a label.
   • Ensure that the file uses comma (,) delimiters.
   • Missing values should be handled before uploading to avoid errors.
3. Features selection.
4. Label shape and color choice.
5. Points pick on plot for detailed view.
   • The application generates a scatter plot where each point represents a reduced-dimension projection of the dataset.
   • Users can click and drag to select a rectangle on the plot, which will reveal detailed feature and MDS-calculated values for the points inside.
   • The plot differentiates labels using user-selected colors and shapes.

License

This project is licensed under the Apache-2.0 License.

Acknowledgements

• University of Konstanz: for MDJS algorithm.