Ctrl+K

Ploting in Pandas

Ploting in Pandas#

Having to load data manually to build a visualization or plot gets cumbersome quickly. In many situations, we might want to work with data in a pandas DataFrame when building a visualization.

The pandas .plot() attribute relies on the matplotlib API to generate plots, so our work with matplotlib will come in handy when we need to customize plots generated using .plot(). And in many cases, the .plot() syntax is similar to matplotlib OO syntax.

We’ll start by working with air quality data.

.plot()#

We can do a quick visual check of the data by passing the entire DataFrame to .plot().

import pandas as pd # import statements
df = pd.read_csv('https://raw.githubusercontent.com/kwaldenphd/elements-of-computing/main/book/data/ch10/air_quality_no2.csv', index_col=0, parse_dates=True) # load data
df # show output
station_antwerp station_paris station_london
datetime
2019-05-07 02:00:00 NaN NaN 23.0
2019-05-07 03:00:00 50.5 25.0 19.0
2019-05-07 04:00:00 45.0 27.7 19.0
2019-05-07 05:00:00 NaN 50.4 16.0
2019-05-07 06:00:00 NaN 61.9 NaN
... ... ... ...
2019-06-20 22:00:00 NaN 21.4 NaN
2019-06-20 23:00:00 NaN 24.9 NaN
2019-06-21 00:00:00 NaN 26.5 NaN
2019-06-21 01:00:00 NaN 21.8 NaN
2019-06-21 02:00:00 NaN 20.0 NaN

1035 rows × 3 columns

df.plot() # plot entire dataframe

<Axes: xlabel='datetime'>
../_images/9fd1cb454e9bb127ca82a74f16d54aa11f8f6775812bd5eb8d9b49a4a224d9d6.png

This isn’t a particularly meaningful visualization, but it shows us how the default for .plot() creates a line for each column with numeric data.

Default settings for Pandas plotting functions:

  • index (in this case DateTime) is the x axis data

  • Numeric columns are the y axis data

  • Default plot type is a line plot

  • Default axis title(s) and legend are pulled from information in the underlying DataFrame

Plotting Specific Columns#

Let’s say we only wanted to plot Paris data. We can plot a specific column in the dataframe using the [" "] selection method before calling .plot().

df["station_paris"].plot() # plot specific column

<Axes: xlabel='datetime'>
../_images/33ab097cf763dfc75e4e9ed02589097f74aae810abc466cc5d69c0ad9b09b950.png

Let’s say we want to visually compare NO2 values measured in London and Paris. We need to specify what column is going to be used for the X axis as well as what column is going to be used for the Y axis.

For this example, a scatterplot will be more effective than a lineplot, so we’ll use .plot.scatter().

df.plot.scatter(x="station_london", y="station_paris", alpha=0.5) # scatterplot for two columns

<Axes: xlabel='station_london', ylabel='station_paris'>
../_images/c1098035df2d12dcfaa269401addf650c44e3a0e4b77f04788cd788b9368d5b8.png

Additional Line Plot Resources#

For more on line plots:

Additional Scatter Plot Resources#

For more on scatter plots:

Recap#

As we’ve already seen Pandas plotting functions incorporate a mix of Pandas syntax and arguments that share similarity with matplotlib.

We’ll see that pattern continue as we explore other plot types and plotting workflows.

Contents