Make lovely multi-scale plots with matplotlib in 3 straightforward steps.

Massive-magnitude outliers, tiny options, and sharp spikes are widespread frustrations to knowledge visualization. All three could make visible particulars illegible by scrunching plot elements into too small an space.

Generally a repair might be had by merely excluding unruly knowledge. When together with such knowledge is chief to a query at hand, making use of a log scale to axes can realign spacing for higher separation amongst decrease magnitude knowledge. This method can solely go to date, nevertheless.

On this article, we’ll check out another choice: zoom plots, which increase a visualization with panels offering magnified views of areas of curiosity.

Zoom plots are generally organized as inserts into the primary plot, however can be mixed as a lattice with the unique plot. We’ll delve into each.

This text supplies a code-oriented tutorial on find out how to use matplotlib with specialised instruments from the outset library to assemble zoom plots. We’ll construct a visualization of rainfall knowledge from Texas made obtainable by Evett et al. by way of the USDA. This knowledge set contains a full yr of rain gauge readings from two close by websites, taken at 15 minute intervals.

The brief period of rain occasions and excessive depth of the heaviest rainfall complicates issues. Throwing a month’s price of Evett et al.’s rainfall knowledge right into a easy line plot of reveals the visualization drawback we’re up towards.

We’ve actually received some work to do to good this up! In our visualization, we’ll give attention to recovering three explicit elements of the information.

1. the little bathe round day 72,
2. the large rainstorm round day 82, and
3. mild precipitation occasions over the course of all the month.

To raised present these particulars, we’ll create a zoom panel for every.

Our plan is laid out, so let’s get into the code 👍

Fetch the rain gauge information by way of the Open Science Framework.

# ----- see appendix for bundle imports
df = pd.read_csv("https://osf.io/6mx3e/obtain")  # obtain knowledge

Right here’s a peek on the knowledge.

+------+-------------+--------------+--------------+------------+-----------+
| Yr | Decimal DOY | NW dew/frost | SW dew/frost | NW precip  | SW precip |
+------+-------------+--------------+--------------+------------+-----------+
| 2019 | 59.73958    | 0            | 0            | 0          | 0         |
| 2019 | 59.74999    | 0            | 0            | 0.06159032 | 0         |
| 2019 | 59.76041    | 0            | 0            | 0          | 0         |
| 2019 | 59.77083    | 0            | 0            | 0.05895544 | 0.0813772 |
| 2019 | 59.78124    | 0            | 0            | 0.05236824 | 0.0757349 |
+ ...  +     ...     +      ...     +      ...     +     ...    +    ...    +

Earlier than shifting on, some minor preparatory chores.

nwls = "NW Lysimetern(35.18817624°N, -102.09791°W)"
swls = "SW Lysimetern(35.18613985°N, -102.0979187°W)"
df[nwls], df[swls] = df["NW precip in mm"], df["SW precip in mm"]
# filter down to simply knowledge from March 2019
march_df = df[np.clip(df["Decimal DOY"], 59, 90) == df["Decimal DOY"]]

Within the code above, we’ve created extra detailed column names and subset the information right down to a single month

Our first plotting step is to initialize an outset.OutsetGrid occasion to handle our latice of magnification plots. This class operates analogously to seaborn’s FacetGrid, which facilitates development of ordinary lattice plots by breaking knowledge throughout axes primarily based on a categorical variable.

OutsetGrid differs from FacetGrid, although, in that along with axes with faceted knowledge it prepares an preliminary “supply” axes containing all knowledge collectively. Additional, OutsetGrid contains instruments to robotically generate “marquee” annotations that present how magnifications correspond to the unique plot. The schematic beneath overviews OutsetGrid’s plotting mannequin.

Getting again to our instance, we’ll assemble an OutsetGrid by offering an inventory of the primary plot areas we need to amplify by way of the knowledgekwarg. Subsequent kwargs present styling and format data.

grid = otst.OutsetGrid(  # initialize axes grid supervisor
knowledge=[
# (x0, y0, x1, y1) regions to outset
(71.6, 0, 72.2, 2),  # little shower around day 72
(59, 0, 90, 0.2),  # all light precipitation events
(81.3, 0, 82.2, 16),  # big rainstorm around day 82
],
x="Time",  # axes label
y="Precipitation (mm)",  # axes label
side=2,  # make subplots huge
col_wrap=2,  # wrap subplots right into a 2x2 grid
# styling for zoom indicator annotations, mentioned later
marqueeplot_kws={"frame_outer_pad": 0, "mark_glyph_kws": {"zorder": 11}},
marqueeplot_source_kws={"zorder": 10, "frame_face_kws": {"zorder": 10}},
)

Right here we’ve specified a wider-than-tall side ratio for subplots and what number of columns we need to have.

Our axes grid is ready up, we’re prepared for the subsequent step.

It’s time to place some content material on our axes.

We are able to use space plots to co-visualize our rain gauges’ readings. (For these unfamiliar, space plots are simply line plots with a fill right down to the x axis.) Making use of a transparency impact will elegantly present the place the gauges agree — and the place they don’t.

We are able to harness matplotlib’s stackplotto attract our overlapped space plots. Though designed to create plots with areas “stacked” on prime of one another, we are able to get overlapped areas by splitting out two calls to the plotter— one for every gauge.

To attract this identical content material throughout all 4 axes of the grid, we are going to use OutsetGrid’s broadcast technique. This technique takes a plotter operate as its first argument then calls it on every axis utilizing any subsequent arguments.

# draw semi-transparent stuffed lineplot on all axes for every lysimeter
for y, coloration in zip([nwls, swls], ["fuchsia", "aquamarine"]):
grid.broadcast(
plt.stackplot,  # plotter
march_df["Decimal DOY"],  # all kwargs beneath forwarded to plotter...
march_df[y],
colours=[color],
labels=[y],
lw=2,
edgecolor=coloration,
alpha=0.4,  # set to 60% clear (alpha 1.0 is non-transparent)
zorder=10,
)

For higher distinction towards background fills, we’ll additionally use broadcast so as to add white underlay across the stackplots.

grid.broadcast(
plt.stackplot,  # plotter
march_df["Decimal DOY"],  # all kwargs beneath forwarded to plotter...
np.most(march_df["SW precip in mm"], march_df["NW precip in mm"]),
colours=["white"],
lw=20,  # thick line width causes protrusion of white border
edgecolor="white",
zorder=9,  # word decrease zorder positions underlay beneath stackplots
)

Right here’s how our plot appears to be like earlier than we transfer on to the subsequent stage.

Wanting good already — we are able to already see magnifications exhibiting up on their correct axes at this stage.

Now it’s time so as to add zoom indicator bins, a.ok.a. outset “marquees,” to indicate how the scales of our auxiliary plots relate to the dimensions of the primary plot.

# draw "marquee' zoom indicators exhibiting correspondences between primary plot
# and outset plots
grid.marqueeplot(equalize_aspect=False)  # enable axes side ratios to differ

Be aware the kwarg handed to permit outset plots to tackle totally different side ratios from the primary plot. This fashion, outset knowledge can absolutely expanded to benefit from all obtainable axes area.

We’re many of the approach there — only a few ending touches left at this level.

Our final enterprise is so as to add a legend and swap out numeric x ticks for correct timestamps.

grid.source_axes.legend(  # add legend to major axes
loc="higher left",
bbox_to_anchor=(0.02, 1.0),  # legend positioning
frameon=True,  # styling: activate legend body
)
# ----- see appendix for code to relabel axes ticks with timestamps

With that, the plot is full.

That’s all there’s to it, a zoom plot in 3 straightforward steps.

We are able to create insets by rearranging the magnification lattice axes into place over the primary axes. Right here’s how, utilizing the outset library’s inset_outsets software.

otst.inset_outsets(
grid,
insets=otst_util.layout_corner_insets(
3,  # three insets
"NW",  # organize in upper-left nook
inset_margin_size=(0.02, 0),  # enable nearer to primary axes bounds
inset_grid_size=(0.67, 0.9),  # develop to take up obtainable area
),
equalize_aspect=False,
)
sns.move_legend(  # transfer legend centered above determine
grid.source_axes, "decrease heart", bbox_to_anchor=(0.5, 1.1), ncol=2
)

On this case, we’ve additionally used outset.util.layout_inset_axes for high-quality tuned management over inset sizing and positioning.

And similar to that, we’ve received three zoom inserts organized within the higher left hand nook.

There’s much more you are able to do with outset.

Along with specific zoom space specification, the outset library additionally supplies a seaborn-like data-oriented API to deduce zoom inserts containing categorical subsets of a dataframe. Intensive styling and format customization choices are additionally obtainable.

Right here’s a peek at some highlights from the library’s gallery…

You’ll be able to be taught extra about utilizing outset within the library’s documentation at https://mmore500.com/outset. Specifically, be sure you take a look at the quickstart information.

Outset might be put in by way of pip as python3 -m pip set up outset.

This tutorial is contributed by me, Matthew Andres Moreno.

I at the moment function a postdoctoral scholar on the College of Michigan, the place my work is supported by the Eric and Wendy Schmidt AI in Science Postdoctoral Fellowship, a Schmidt Futures program.

My appointment is break up between the college’s Ecology and Evolutionary Biology Division, the Middle for the Research of Complexity, and the Michigan Institute for Information Science.

Discover me on Twitter as @MorenoMatthewA and on GitHub as @mmore500.

disclosure: I’m the writer of the outset library.

Evett, Steven R.; Marek, Gary W.; Copeland, Karen S.; Howell, Terry A. Sr.; Colaizzi, Paul D.; Brauer, David Ok.; Ruthardt, Brice B. (2023). Evapotranspiration, Irrigation, Dew/frost — Water Stability Information for The Bushland, Texas Soybean Datasets. Ag Information Commons. https://doi.org/10.15482/USDA.ADC/1528713. Accessed 2023–12–26.

J. D. Hunter, “Matplotlib: A 2D Graphics Setting”, Computing in Science & Engineering, vol. 9, no. 3, pp. 90–95, 2007. https://doi.org/10.1109/MCSE.2007.55

Marek, G. W., Evett, S. R., Colaizzi, P. D., & Brauer, D. Ok. (2021). Preliminary crop coefficients for late planted short-season soybean: Texas Excessive Plains. Agrosystems, Geosciences & Setting, 4(2). https://doi.org/10.1002/agg2.20177

Information buildings for statistical computing in python, McKinney, Proceedings of the ninth Python in Science Convention, Quantity 445, 2010. https://doi.org/ 10.25080/Majora-92bf1922–00a

Waskom, M. L., (2021). seaborn: statistical knowledge visualization. Journal of Open Supply Software program, 6(60), 3021, https://doi.org/10.21105/joss.03021.

You’ll find all the code as a gist right here and as a pocket book right here.

To put in dependencies for this train,

python3 -m pip set up 
matplotlib `# ==3.8.2`
numpy `# ==1.26.2` 
outset `# ==0.1.6` 
opytional `# ==0.1.0` 
pandas `# ==2.1.3` 
seaborn `# ==0.13.0`

All pictures are works of the writer.