Path plots

In [1]:

import pandas as pd

from plotnine import (
    ggplot,
    aes,
    geom_path,
    geom_line,
    labs,
    scale_color_continuous,
    element_text,
    theme,
)
from plotnine.data import economics

geom_path() connects the observations in the order in which they appear in the data, this is different from geom_line() which connects observations in order of the variable on the x axis.

In [2]:

economics.head(10)  # notice the rows are ordered by date

	date	pce	pop	psavert	uempmed	unemploy
0	1967-07-01	507.4	198712	12.5	4.5	2944
1	1967-08-01	510.5	198911	12.5	4.7	2945
2	1967-09-01	516.3	199113	11.7	4.6	2958
3	1967-10-01	512.9	199311	12.5	4.9	3143
4	1967-11-01	518.1	199498	12.5	4.7	3066
5	1967-12-01	525.8	199657	12.1	4.8	3018
6	1968-01-01	531.5	199808	11.7	5.1	2878
7	1968-02-01	534.2	199920	12.2	4.5	3001
8	1968-03-01	544.9	200056	11.6	4.1	2877
9	1968-04-01	544.6	200208	12.2	4.6	2709

Because the data is in date order geom_path() (in pint) produces the same result as geom_line() (in black):

In [3]:

(
    ggplot(economics, aes(x="date", y="unemploy"))
    + geom_line(size=5)  # plot geom_line as the first layer
    + geom_path(
        colour="#ff69b4",  # plot a path - colour pink
        size=1,
    )
    + labs(x="date", y="unemployment (,000)")  # label x & y-axis
)

Plotting unemployment (scaled by population) versus savings rate shows how geom_path() differs from geom_line(). Because geom_path() connects the observations in the order in which they appear in the data, this line is like a “journey through time”:

In [4]:

(
    ggplot(economics, aes(x="unemploy/pop", y="psavert"))
    + geom_path()  # plot geom path
    + labs(x="scaled unemployment", y="saving rate")  # label x & y-axis
)

Comparing geom_line() (black) to geom_path() (pink) shows how these two plots differ in what they can show: geom_path() shows the savings rate has gone down over time, which is not evident with geom_path().

In [5]:

(
    ggplot(economics, aes(x="unemploy/pop", y="psavert"))
    + geom_path(
        colour="#ff69b4",  # plot geom_path as the first layer - colour pink
        alpha=0.5,  # line transparency
        size=2.5,
    )  # line thickness
    + geom_line()  # layer geom_line
    + labs(x="scaled unemployment", y="saving rate")  # label x & y-axis
)

The geom_path can be easier to interpret if time is coloured in. First convert time to a number, and use this number to colour the path:

In [6]:

# convert date to a number
economics["date_as_number"] = pd.to_numeric(economics["date"])

In [7]:

# inspect new column
economics.head()

	date	pce	pop	psavert	uempmed	unemploy	date_as_number
0	1967-07-01	507.4	198712	12.5	4.5	2944	-79056000000000000
1	1967-08-01	510.5	198911	12.5	4.7	2945	-76377600000000000
2	1967-09-01	516.3	199113	11.7	4.6	2958	-73699200000000000
3	1967-10-01	512.9	199311	12.5	4.9	3143	-71107200000000000
4	1967-11-01	518.1	199498	12.5	4.7	3066	-68428800000000000

The path is coloured such that it changes with time using the command aes(colour='date_as_number') within geom_path().

In [8]:

# input
legend_breaks = [
    -79056000000000000,
    709948800000000000,
    1401580800000000000,
]  # used to modify colour-graded legend
legend_labels = ["1967-Jul", "1992-Jun", "2014-Jun"]

# plot
(
    ggplot(economics, aes(x="unemploy/pop", y="psavert"))
    + geom_path(
        aes(colour="date_as_number")
    )  # colour geom_path using time variable "date_as_number"
    + labs(x="scaled unemployment", y="saving rate")
    + scale_color_continuous(
        breaks=legend_breaks,  # set legend breaks (where labels will appear)
        labels=legend_labels,
    )  # set labels on legend
    + theme(legend_title=element_text(text="Date"))  # set title of legend
)