Code and slides to accompany the PyGotham webinars: https://2020.pygotham.tv/talks/visualization-with-seaborn/ by Data For Science.
Seaborn is a visualization package that builds on top of matplotlib and pandas to provides a simple, functional, interface that is capable of generating sophisticated and beautiful visualizations. In this lecture we will provide a systematic overview of the way in which seaborn is structured, how it can be used for data exploration and to produce publication ready figures and visualizations.
- Basic function structure - seaborn uses a simple and systematic structure for its function calls, making it easy to quickly experiment with different plot types.
- Axes level vs Figure level functions - axes level functions generate a single plot, while figure level functions have the ability to automatically produce several subplots within the same figure. We will cover how to take full advantage of both.
- Relationship plots - relationship plot are useful to plot relationships between continuous variables
- Categorical plots - categorical plots have at least one categorical feature that can be used to slice the data in various ways
- Using FaceGrid - FaceGrid is a seaborn feature that allows us to define customized plotting functions to generate each subplot of a figure while taking advantage of seaborns functionality to slice and aggregate the data
- Customizing seaborn with matplotlib - All figures generated by seaborn use matplot objects. We will learn how to access these objects and take advantage of matplotlib to further customize them.