Created: 2022-11-01 13:11:07 Tags: #ML #DataScience #classification_metric #ml_metrics
# Note
Accuracy is one of the metrics for classification model evaluation.
Informally, accuracy is a fraction of predictions our model got right:
$Accuracy = \frac{number_of _correct_predictions}{total_number_predictions}$
Also, we can define accuracy in Confusion Matrix terms:
$Accuracy = \frac{TP + TN}{TP + TN + FP + FN}$
# Advantages
- Easy for understanding
- Easy for calculation
# Disadvantages
- Works badly with imbalabced class problem. For imbalanced classes you can use Precision, Recall and F measure to evaluate model performance.