museotoolbox.ai.SuperLearner.save_cm_from_cv

SuperLearner.save_cm_from_cv(savePath, prefix='', header=True, n_jobs=1)[source]

Save each confusion matrix (csv format) from cross-validation.

For each matrix, will save as header :

  • The number of training samples per class,

  • The F1-score per class,

  • Overall Accuracy,

  • Kappa.

Example of confusion matrix saved as csv :

# Training samples : 90,80

# F1 : 91.89,90.32

# OA : 91.18

# Kappa : 82.23

85

5

10

70

  • In X (columns) : prediction (95 predicted labels for class 1).

  • In Y (lines) : reference (90 labels from class 1).

Parameters

  • savePath (str.) – The path where to save the different csv. If not exists, will be created

  • prefix (str, default ''.) – If prefix, will add this prefix before the csv name (i.e. 0.csv)

  • header (boolean, default True.) – If True, will save F1, OA, Kappa and number of training samples. If False, will only save confusion matrix

Returns

Return type

None

Examples

After having learned with museotoolbox.ai.SuperLearner :

>>> SL.saveCMFromCV('/tmp/testMTB/',prefix='RS50_')
[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.
[Parallel(n_jobs=-1)]: Done  10 out of  10 | eSLsed:    3.4s finished
>>> np.loadtxt('/tmp/testMTB/RS50_0.csv')
array([[85,  5],
[10, 70]])