CONFIDENCE CIRCLE ON CORRESPONDENCE ANALYSIS USING SINGULAR VALUE DECOMPOSITION

• Main Author: Dwi Kurniawati, Darmayanti
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77281
• Institution: Institut Teknologi Bandung

Correspondence analysis is one of the method of multivariate data analysis to find special pattern in a data. This method is useful for visualizing qualitative data in a more attractively graphical presentation so it can be understood easily. Correspondence analysis algorithm will produce two or three euclidean subspaces, and will project all rows and columns profile in those euclidean subspaces. The dimension reduction technique used in this thesis for determining the euclidean subspaces is the Singular Value Decomposition (SVD). In this thesis, Correspondency Analysis will be used to discover the tendency of Graduates in choosing their jobs based on their GPA using tracer study data. In the last step, the correspondency analysis maps will also contain confidence circle. These confidence circles will show how close are several coordinates in different categories with each other. Also it will aids in highlighting some categories that doesn't have significant influence to the correspondency.