pymatviz 0.4.4

A toolkit for visualizations in materials informatics

pymatviz

A toolkit for visualizations in materials informatics.

Note: This project is not associated with or endorsed by pymatgen, but aims to complement it by adding additional plotting functionality.

Installation

pip install pymatviz

Usage

Check out the Jupyter notebooks under examples/ to learn how to use pymatviz.

Elements

See pymatviz/elements.py. Heat maps of the periodic table can be plotted both with matplotlib and plotly. Latter supports displaying additional data on hover or full interactivity through Dash.

ptable_heatmap(compositions, log=True)	ptable_heatmap_ratio(comps_a, comps_b)
ptable_heatmap_plotly(atomic_masses)	ptable_heatmap_plotly(compositions)

Sunburst

See pymatviz/sunburst.py.

spacegroup_sunburst([65, 134, 225, ...])	spacegroup_sunburst(["C2/m", "P-43m", "Fm-3m", ...])

Sankey

See pymatviz/sankey.py.

sankey_from_2_df_cols(df_perovskites)	sankey_from_2_df_cols(df_rand_ints)

Structure

See pymatviz/struct_viz.py. Currently structure plotting is only supported with matplotlib in 2d. 3d interactive plots (probably with plotly) are on the road map.

plot_structure_2d(mp_19017)	plot_structure_2d(mp_12712)

Histograms

See pymatviz/histograms.py.

spacegroup_hist([65, 134, 225, ...])	spacegroup_hist(["C2/m", "P-43m", "Fm-3m", ...])
residual_hist(y_true, y_pred)	hist_elemental_prevalence(compositions, log=True, bar_values='count')

Parity Plots

See pymatviz/parity.py.

density_scatter(xs, ys, ...)	density_scatter_with_hist(xs, ys, ...)
density_hexbin(xs, ys, ...)	density_hexbin_with_hist(xs, ys, ...)
scatter_with_err_bar(xs, ys, yerr, ...)	residual_vs_actual(y_true, y_pred, ...)

Uncertainty Calibration

See pymatviz/quantile.py.

qq_gaussian(y_true, y_pred, y_std)	qq_gaussian(y_true, y_pred, y_std: dict)

Ranking

See pymatviz/ranking.py.

err_decay(y_true, y_pred, y_std)	err_decay(y_true, y_pred, y_std: dict)

Cumulative Error and Residual

See pymatviz/cumulative.py.

cum_err(preds, targets)	cum_res(preds, targets)

Classification Metrics

See pymatviz/relevance.py.

roc_curve(targets, proba_pos)	precision_recall_curve(targets, proba_pos)

Correlation

See pymatviz/correlation.py.

marchenko_pastur(corr_mat, gamma=ncols/nrows)	marchenko_pastur(corr_mat_significant_eval, gamma=ncols/nrows)

Glossary

1. Residual y_res = y_true - y_pred: The difference between ground truth target and model prediction.
2. Error y_err = abs(y_true - y_pred): Absolute error between target and model prediction.
3. Uncertainty y_std: The model's estimate for its error, i.e. how much the model thinks its prediction can be trusted. (std for standard deviation.)

Usage on Colab

For the time being, Google Colab only supports Python 3.7. pymatviz uses Python 3.8 features like self-documenting f-strings. You may still be able to use pymatviz on Colab by cloning the repo and patching the source code in place as shown here.