scanpy.metrics.confusion_matrix

scanpy.metrics.confusion_matrix(orig, new, data=None, *, normalize=True)

Given an original and new set of labels, create a labelled confusion matrix.

Parameters orig and new can either be entries in data or categorical arrays of the same size.

Parameters

orig : Series | ndarray | SequenceUnion[Series, ndarray, Sequence]

Original labels.

new : Series | ndarray | SequenceUnion[Series, ndarray, Sequence]

New labels.

data : DataFrame | NoneOptional[DataFrame] (default: None)

Optional dataframe to fill entries from.

normalize : boolbool (default: True)

Should the confusion matrix be normalized?

Examples

import scanpy as sc; import seaborn as sns
pbmc = sc.datasets.pbmc68k_reduced()
cmtx = sc.metrics.confusion_matrix("bulk_labels", "louvain", pbmc.obs)
sns.heatmap(cmtx)

Return type

DataFrameDataFrame