Technical and Research Skills: Techniques

Exploratory Data Analysis

1 day

There is a pressing need for techniques to help us understand high multivariate data possibly with large numbers of cases and/or variety of data types. This half-day workshop aims to give participants experience of techniques in Exploratory Data Analysis (EDA) – a framework for seeing hypotheses which complements conventional statistical testing (Confirmatory Data Analysis).

Data visualisation – a key component of EDA – shows us a ‘big picture’ of a dataset to discover its internal structure. John Turkey, who originated EDA, described data visualisation as having its greatest value ‘when it forces us to notice what we never expected to see’.

On completing the course, participants will:

• be aware of the EDA framework as complementary to confirmatory analysis;
• have understanding of how Open Source data visualisation tolls such as ‘GGobi’, ‘Mondrian’ and selected graphical options in the ‘R’ language can depict a variety of data sets;
• have experience of newer techniques like Parallel Co-ordinates and Mosaic plots for depicting data sets with multiple variables and/or large numbers of cases to find relationships, clusters and outliers.

Provider: UGC
Target Group: RS & PGR