Exploratory Data Analysis in R

Four graphical techniques you can use to quickly explore your data

This example-based course introduces exploratory data analysis (EDA) using R. A primary objective is to apply graphical EDA techniques to representative data sets using the RStudio platform.

I have incorporated datasets from the NIST/SEMATECH e-Handbook of Statistical Methods into this course and adopted their fundamental approach of Exploratory Data Analysis.

We use scatter plots to examine relationships between two variables, determine if there is a linear or non-linear relationship, analyze variations of the dependent variable, and determine if there are outliers in the dataset.

Of course, we need to remember that causality implies association and that association does NOT imply causality.

We will summarise the distribution of a dataset graphically using histograms. This tool can quickly show us the location and spread of the data, and give us a good indication if the data follows a normal distribution, is skewed, has multiple modes or outliers.

An underused, complementary technique to histograms is the probability plot. We will construct probability plots by plotting the data against a theoretical normal distribution. If the data follows a normal distribution, the plot will form a straight line. We will use the normal probability plot to assess whether or not our examples follow a normal distribution.

Finally, we will use box plots to view the variation between different groups within the data.

Aside from scatterplots, most spreadsheet programs do not support these methods, so learning how to do this fundamental analysis in R can improve your ability to explore your data.