Exploratory Data Analysis

Department of

Computational Perception

Home – Mission – Teaching – People – Research – Media – Impressum

Special Topics: Exploratory Data Analysis

344.044, KV, 2h (3 ETCS), Winter Term 2016/17

Assoc.-Prof. Dr. Markus Schedl

Time: Tuesday, 13:45 - 15:15

Start: October 11, 2016

Location: S2 054

The lecture will be held in English.

Goals

This course gives an introduction to techniques, tools, and applications used in Visual Data Mining and methods underlying Exploratory Data Analysis. Special emphasis is given to linear and non-linear techniques for high-dimensional data projection, clustering (unsupervised learning), and related information visualization methods. Students will also be given the opportunity to implement and test these techniques in a pratical exercise.

The main topics covered include:

Basics: motivation for EDA, categorization of EDA techniques, different kinds of input data, similarity functions
Spectral Methods: Principal Components Analysis (PCA), Multi-dimensional Scaling (MDS), Sammon's Mapping, various fitness functions
Self-Organizing Maps: SOM, GHSOM, Aligned SOM
Visualizing SOMs: SOM grid, Music Description Map (MDM), Bar Plots, Chernoff's Faces, U-Matrix, Distance Matrix, Smoothed Data Histogram (SDH), Component Planes
Graph-based Methods: Locally Linear Embedding (LLE), Isomap, Laplacian Eigenmaps
Probabilistic Models: Stochastic Neighbor Embedding (SNE) and t-Distributed Stochastic Neighbor Embedding (t-SNE)
Information Visualization: TreeMaps, InterRing/Sunburst, Hyperbolic Browser, Time-Series Visualization

Schedule

The detailed schedule can be found on the course's KUSSS web page.

last edited by ms on Sep 6, 2016