Which metric is LEAST appropriate for evaluating recommendation ranking quality?

The correct answer is D. Accuracy. Recommendation systems are fundamentally ranking problems: the goal is to surface the most relevant items at the top of a list. NDCG (Normalized Discounted Cumulative Gain), MAP (Mean Average Precision), and Precision@K all explicitly reward models for placing relevant items high

NVIDIA Certified Associate (NCA) Core AI Concepts

Question

Options

ANDCG
BMAP
CPrecision@K
DAccuracy

How the community answered

(28 responses)

A
4% (1)
B
4% (1)
C
7% (2)
D
86% (24)

Explanation

Recommendation systems are fundamentally ranking problems: the goal is to surface the most relevant items at the top of a list. NDCG (Normalized Discounted Cumulative Gain), MAP (Mean Average Precision), and Precision@K all explicitly reward models for placing relevant items higher in the ranked list. Accuracy, by contrast, treats every position equally (correct or incorrect, rank 1 or rank 100 is the same) and is designed for flat classification tasks. In a recommendation context, accuracy ignores ranking order entirely, making it a poor proxy for whether the system actually surfaces the best items first.

Topics

#Recommendation Systems#Evaluation Metrics#Ranking Algorithms#Machine Learning

Community Discussion

No community discussion yet for this question.

Full NCA-AIIO Practice