Model Performance Metrics

This section will cover model performance metrics, including common metrics for classification, regression, ranking, image object detection, image semantic segmentation, NLP semantic segmentation, NLP classification, NLP named entity recognition, clustering, forecasting, and other models. Check back for updates as we add chapters!

Tabular

Regression

• MAE
• MAPE
• RMSE
• MSE

Classification

• AUC
• LogLoss
• Accuracy
• Precision
• Recall
• Accuracy
• F1

Ranking

• NDCG
• DCG
• Recall@
• Prevision@

Image

Classification

• Precision
• Recall
• Accuracy
• F1

Object Detection (bounding box)

• Average IOU
• Map
• AP

Segmentation

• Average IOU
• AP
• MAP

Depth Estimation

• Absolute relative difference (Abs-Rel)
• Square Relative Difference (SQ-Rel)
• Absolute Relative Distance (ARD)
• RMSE

Video Classification

• Accuracy
• Precision
• F1
• Recall
• MAP
• IOU

Zero Shot Image Classification

• Top K Accuracy
• Seen Class Accuracy
• Novel Class Accuracy

NLP

Conversational

• Blue Score

Fill Mask

• Perplexity
• KL Divergence
• Cross Entropy

Question Answering

• F1
• Exact Match

Sentence Similarity

• Euclidean Distance (Embedding)
• Cosine Distance (Embedding)
• Reciprocal Rank
• NDCG

Summarization

• ROUGE Score

Text Classification

• Accuracy
• Precision
• Recall
• F1

Token Classification (NER, POS)

• Accuracy
• Precision
• Recall
• F1

Translation

• Bleu

Text Generation

• Perplexity
• KL Divergence
• Cross Entropy
• PSI