Class Prediction Act

A Class Prediction Act is a prediction act that assigns class membership to an input item from a finite class set.

• AKA: Classification Act, Categorization Act, Class Assignment Act, Class Decision Act, Class Labeling Act.
• Context:
  • It can typically produce a predicted class label with confidence scores.
  • It can instantiate a Class Prediction Task through classifier execution.
  • It can often generate class prediction decisions based on learned patterns.
  • It can range from being a Binary Class Prediction Act to being a Multiclass Prediction Act, depending on its class set size.
  • It can range from being a Correct Class Prediction Act to being an Incorrect Class Prediction Act, depending on its prediction accuracy.
  • It can range from being a Manual Class Prediction Act to being an Automated Class Prediction Act, depending on its execution method.
  • It can incorporate decision thresholds for binary class prediction.
  • It can support classifier performance evaluation through confusion matrix generation.
  • It can enable prediction explanation through feature importance analysis.
  • ...
• Example(s):
  • Binary Class Prediction Acts, such as:
    • Spam Email Classification Act that predicts spam/not-spam.
    • Disease Diagnosis Prediction Act that predicts present/absent.
    • Credit Default Classification Act that predicts default/non-default.
    • Fraud Detection Act that predicts fraudulent/legitimate.
  • Multiclass Prediction Acts, such as:
    • Document Topic Classification Act for news categorization.
    • Image Object Classification Act identifying multiple object types.
    • Sentiment Polarity Classification Act assigning positive/negative/neutral.
  • Multilabel Class Prediction Acts, such as:
    • Tag Assignment Act for document tagging.
    • Multi-Disease Diagnosis Act predicting multiple conditions.
  • Hierarchical Class Prediction Acts for taxonomic classification.
  • Probabilistic Class Prediction Acts outputting class probabilities.
  • ...
• Counter-Example(s):
  • Regression Prediction Act, which outputs continuous values rather than classes.
  • Clustering Act, which groups without predefined class labels.
  • Ranking Prediction Act, which orders items rather than classifying.
  • Anomaly Detection Act, which identifies outliers without class assignment.
  • Dimensionality Reduction Act, which transforms without predicting.
• See: Class Prediction Task, Classification Algorithm, Predicted Class, Classification Error, Confusion Matrix, Type I Error, Type II Error, ROC Analysis.

---

---

References

2002

• (Gabor Melli, 2002) ⇒ Gabor Melli. (2002). “PredictionWorks' Data Mining Glossary." PredictionWorks.
  • Classification/Classifier: The act of labeling a test case into one of a finite number of output classes. A model that classifies is sometimes referred to as a "classifier". Commonly a classifer's performance is measured by its ability to correctly label unseen test cases, that is its "accuracy". Inversely a classifier's performance may be measured by its "error rate". A more detailed insight into a classifier's performance is given by the Confusion Matrix structure because it captures how well the classifier predicts each of the available classes. If a Cost-Benefit Matrix is available then the classifier's performance is measured by the product of the Confusion and Cost-Benefit matrices.

---