This is a Python module that contains some useful data visualization tools.

Table of contents:

• Python plotting utilities: plot_utils
  • 1. Installation - Note:
  • 2. API documentations
  • 3. Current functionalities
    • 3.1. Visualizing one column of data
    • 3.2. Visualizing two columns of data ([example])
    • 3.3. Visualizing multiple columns of data
    • 3.4. Map plotting
    • 3.5. Time series plotting
    • 3.6. Miscellaneous
  • 4. Gallery
    • 4.1. One column of data
    • 4.2. Two columns of data
    • 4.3. Multiple columns of data
      • 4.3.1. 3D histograms
      • 4.3.2. Multi-histogram plots
      • 4.3.3. Violin plots
      • 4.3.4. Correlation matrix
      • 4.3.5. Count missing values
    • 4.4. Choropleth maps (a.k.a., "heat maps")
      • 4.4.1. State-level choropleth maps for the US
      • 4.4.2. County-level choropleth maps for the US
    • 4.5. Time series plotting
      • 4.5.1. Single time series
      • 4.5.2. Multiple time series
    • 4.6. Miscellaneous
      • 4.6.1. get_colors() function
      • 4.6.2. get_linespecs() function
      • 4.6.3. Plot with error bounds
      • 4.6.4. Visualize cross-validation scores
  • 5. Dependencies
  • 6. Testing
  • 7. Aesthetics
  • 8. References
  • 9. Copyright and license

In the terminal, run the following command:

pip install plot-utils

If you run into the following issue on Mac OS X (or macOS) when importing plot_utils:

RuntimeError: Python is not installed as a framework. The Mac OS X backend will not be able to function correctly if Python is not installed as a framework.

please follow this solution to fix the issue: https://stackoverflow.com/a/21789908/8892243

All API documentations are on: https://python-plot-utils.readthedocs.io/en/stable/index.html.

Current functionalities include (for full list, use print(plot_utils.__doc__)):

• Pie chart: proportions of distinct values in an array, more convenient than matplotlib's pie() function [doc], [example]
• Discrete histogram: counts of distinct values in an array [doc], [example]

• "Bin-and-mean" plot: for two continuous variables [doc]
• Category mean: for a categorical variable and a continuous variable [doc]
• Positive rate: for a categorical variable and a binary categorical variable [doc]
• Contingency table: for two categorical variables [doc]

• 3D histograms: distributions of multiple variables [doc], [example]
• Multiple histograms: distribution of multiple variables [doc], [example]
• Violin plot: distribution of multiple variables [doc], [example]
• Correlation matrix: correlation between each columns of a dataset [doc], [example]
  • and the one-to-one scatter plots for the variables within the dataset [doc]
• Count missing values: how many missing values are there in each column of the dataset [doc], [example]

• Choropleth map (a.k.a., "heat map") of the United States, on both the state and county level [doc], [example]

• Plot single time series: [doc], [example]
• Plot multiple time series: [doc], [example]
• Fill time series with error bounds: [doc], [example]

• A get_colors() function that conveniently queries different color palettes [doc], [example]
• A get_linespecs() function that generates distinct color/linestyle/linewidth combinations for plotting many lines together [doc], [example]
• Two helper classes: Color and Multiple_Colors, which make querying and displaying colors more easily [doc], [example]
• Plotting with error bounds, which displays error bounds as shaded areas [doc], [example]
• trim_img(), which trims the white margins of the specified image file(s) [doc]
• pad_img(), which pads image(s) with margins so that they meet a specified aspect ratio [doc]
• plot_ranking(), which ranks a series of values and shows them as a bar chart [doc], [example]
• visualize_cv_scores(), which visualizes cross-validation training/evaluation scores [doc], [example]

Using the "Titanic dataset" as example:

import pandas as pd
import plot_utils as pu
titanic = pd.read_csv('./examples/datasets/titanic3.csv')
pu.piechart(titanic['survived'], title='Suvived')
pu.discrete_histogram(titanic['pclass'], xlabel='Passenger ticket class')

Piechart: [doc], [example]

Discrete histogram: [doc], [example]

Using the "Titanic dataset" as example:

titanic = pd.read_csv('./examples/datasets/titanic3.csv')
titanic.rename(index=str, columns={'pclass':'ticket_class'}, inplace=True)
titanic['embarked'] = titanic['embarked'].map(
    {'S':'Southampton', 'C':'Cherbourg', 'Q':'Queenstown'}
)

pu.bin_and_mean(titanic['age'], titanic['fare'], figsize=(5, 3))
pu.category_means(titanic['ticket_class'], titanic['fare'])
pu.positive_rate(titanic['ticket_class'], titanic['survived'], figsize=(5, 2))
pu.contingency_table(titanic['ticket_class'], titanic['embarked'], dropna=True, rot=0)

["bin-and-mean" doc] ["category means" doc] ["positive rate" doc] ["contingency table" doc] [All examples]

Useful for comparing multiple distributions:

iris = pd.read_csv('./examples/datasets/iris.csv')
pu.histogram3d(iris[['petal_width', 'petal_length', 'sepal_width', 'sepal_length']])

[doc], [example]

Another useful tool to compare multiple distributions:

pu.hist_multi(iris[['sepal_length', 'sepal_width', 'petal_length', 'petal_width']]);

[doc], [example]

To compare multiple distributions:

pu.violin_plot(iris[['petal_width', 'petal_length', 'sepal_width', 'sepal_length']])

[doc], [example]

iris = pd.read_csv('./examples/datasets/iris.csv')
pu.plot_correlation(iris, scatter_plots=True)

The first figure shows the correlation (or "sample covariance") between each column. The second figure shows the scatter plots between each pair of columns.

[doc], [example]

To see how much data is missing in each column of a data set:

titanic = pd.read_csv('./examples/datasets/titanic3.csv')
pu.missing_value_counts(titanic)

[doc], [example]

Each bar corresponds to a column in titanic, and the numbers atop are the missing data counts for the corresponding column.

pu.choropleth_map_state(state_level_data)  # `state_level_data`: dict, pandas.DataFrame, or pandas.Series

[doc], [example]

pu.choropleth_map_county(county_level_data)  # `county_level_data`: dict, pandas.DataFrame, or pandas.Series

[doc], [example]

df = pd.read_csv('./examples/datasets/Unemployment_rate_1976-2017.csv', index_col=0)
pu.plot_timeseries(df['CA'], ylabel='Unit: %', title='Unemployment rate, California')

[doc], [example]

pu.plot_multiple_timeseries(df, ylabel='Unemployment rate [%]', ncol_legend=10)

[doc], [example]

Easy querying of distinguishable color palettes.

colors = pu.get_colors(color_scheme='tab10', N=10)  # `colors`: a list containing 10 colors
pu.Multiple_Colors(colors).show()  # show colors as a palette

[doc], [example]

Easy querying of distinguishable line specs.

line_specs = pu.get_linespecs(color_scheme='bw',range_linewidth=[3,8],priority='linewidth')
pu.linespecs_demo(line_specs)

[doc], [example]

[get_linespecs() doc]

Plots data and error bounds on the same graph.

pu.plot_with_error_bounds(x, y, y_upper_bound, y_lower_bound)

[doc], [example]

pu.visualize_cv_scores(n_folds=4, cv_scores=[0.675, 0.702, 0.689, 0.653], holdout_score=0.6745);

[doc], [example]

• Python 2.7 or 3.5+
• matplotlib 1.5.0+, or 2.0.0+ (Version 2.1.0+ is strongly recommended.)
• numpy: 1.11.0+
• scipy: 0.19.0+
• pandas: 0.20.0+
• cycler: 0.10.0+
• matplotlib/basemap: 1.0.7 (only if you want to plot the two choropleth maps)
• PIL (only if you want to use trim_img() or pad_img())

To test plot_utils, run the ipython notebooks in the examples folder.

The aesthetics of of the plot_utils module are matplotlib-styled by default, but it doesn't mean that you can't use your favorite styles in seaborn or ggplot2.

Unlike some plotting packages that enforces their own styles and restrict users from customizing, users of this module can adjust the figure styles freely: either from within matplotlib (https://matplotlib.org/devdocs/gallery/style_sheets/style_sheets_reference.html), or import seaborn and let seaborn take care of everything.

I did not built every function of this module entirely from scratch. I documented the sources that I referenced in the documentation of the corresponding functions.

(c) 2017-2023, jsh9

License: GPL v3.0