Note
Go to the end to download the full example code
Confusion Matrix Report#
This notebook provides an overview for using and understanding the Confusion Matrix Report check for NLP tasks.
Structure:
What is the Confusion Matrix Report?#
The ConfusionMatrixReport produces a confusion matrix visualization which summarizes the
performance of the model. The confusion matrix contains the TP (true positive), FP (false positive),
TN (true negative) and FN (false negative), from which we can derive the relevant metrics,
such as accuracy, precision, recall etc. (confusion matrix).
Generate data & model#
from deepchecks.nlp import TextData
from deepchecks.nlp.checks import ConfusionMatrixReport
from deepchecks.nlp.datasets.classification.tweet_emotion import load_data, load_precalculated_predictions
tweets_data = load_data(data_format='DataFrame', as_train_test=False)
tweets_dataset = TextData(tweets_data.text, label=tweets_data['label'],
task_type='text_classification')
predictions = load_precalculated_predictions(as_train_test=False)
include_properties and include_embeddings are incompatible with data_format="Dataframe". loading only original text data.
Run the check#
check = ConfusionMatrixReport()
result = check.run(tweets_dataset, predictions=predictions)
result.show()
Define a condition#
We can define our check a condition that will validate if all the misclassified
cells/samples in the confusion matrix is below a certain threshold. Using the
misclassified_samples_threshold argument, we can specify what percentage of the total samples
our condition should use to validate the misclassified cells.
# Let's add a condition and re-run the check:
check = ConfusionMatrixReport()
check.add_condition_misclassified_samples_lower_than_condition(misclassified_samples_threshold=0.1)
result = check.run(tweets_dataset, predictions=predictions)
result.show()
#%%
Total running time of the script: (0 minutes 0.161 seconds)