Types of data and data types

Please note that this page appears in both the chapters “Descriptive Statistics” and “Pandas Basics”. This is because the topic is relevant to both the general understanding of data and the specific implementation in Pandas.

Learning objectives

After working through this topic, you should be able to:

• determine the correct description of a given data column

  • nominal

  • cardinal

  • ordinal

• explain how these categories map into Pandas data types:

  • unordered categorical

  • floating point

  • integer

  • ordered categorical

Materials

Video with English subtitles:

Download the slides.

Video with German subtitles:

(turn subtitles on in the bottom right corner of the video)

Decision tree

This is the decision tree for the “correct” form of data. Just because a variable arrives as a number, it does not mean that you should think about it as numerical data. Very often this happens when categories are encoded as numbers (e.g., 0, 1, 2 meaning \([0, 30,000)\), \([30,000, 60,000)\), \([60,000, \infty)\), which would be described in some metadata).

flowchart LR A -->|"`Yes`"| C("`Numerical?`") A(Ordered?) -->|No| B("`Unordered Categorical`") C -->|"`Yes`"| E("`Continuous?`") C -->|"`No`"| D("`Ordered Categorical`") E -->|"`Yes`"| G("`Floating point`") E -->|"`No`"| F("`Integer`") style A fill:#FFE5B4 style C fill:#FFE5B4 style E fill:#FFE5B4

Note that for numerical data, you still have to decide whether a variable is measured on a cardinal or ordinal scale. Both are possible for continuous or discrete data, it is not embedded in the Pandas data type.