Exploratory data analysis

When your data sits in a SAS dataset or other database, it’s difficult to observe much about it beyond its size and the types of variables it contains. In this tutorial, we’ll learn about summarizing and visualizing your data. Sometimes, we refer to the techniques covered in this tutorial as exploratory data analysis or EDA.

EDA is all about learning the structure of a dataset through a series of numerical and graphical techniques. When you do EDA, you’ll generate summaries of your data that help you understand the distributions of variables, uncover relationship, and find outliers. EDA can also generate questions that will help inform subsequent analysis.

Learning objectives

Visualize categorical and numerical data using appropriate graphics.
Create graphical representations of multiple variables.
Describe the structure revealed by graphics in the language of distributions.
Use statistics to summarize important aspects of data.
Identify unusual observations

Lessons

Exploring single variables

Create graphical and numerical summaries to understand the marginal distributions of single variables

Categorical variables
- Frequencies & relative frequency tables
- Bar charts and pie charts
Numerical variables
- Measures of center
- Measures of spread (variability)
- Distribution shape with histograms
- Outlier detection with box plots
- Transformations to reduce skew in graphics

Exploring relationships between two variables

Create graphics that display and calculate numeric measures that summarize the relationship between multiple variables

Categorical variable x categorical variable
- Contingency tables
- Segmented bar charts and mosaic plots
Numerical variable x numerical variable
- Scatterplot
Categorical variable x numerical variable
- By group measures of center and variability
- Paneled histograms & box plots

Measures of association between two variables

Calculate numeric measures that summarize the relationship between multiple variables

Categorical variable x categorical variable
- Relative risk & risk difference
Numerical variable x numerical variable
- Correlation coefficient
Categorical variable x numerical variable
- Mean differences