What does this research mean for the field?

Hybrid learned cardinality estimation methods, particularly FactorJoin, achieve the highest accuracy on data warehouse workloads with star and snowflake schemas, while data-driven methods like FLAT and BayesCard provide the optimal balance of accuracy and inference speed. Novelty: ClaimNovelty.CONFIRMATORY. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This paper aims to evaluate the effectiveness of learned cardinality estimation methods on various data warehouse schemas.

May 20, 2026Journal of Computer Technology and Applied MathematicsOpen Access

Benchmarking Learned Cardinality Estimation Techniques for Analytical Query Processing in Data Warehouses

View Full Paper

Ask AI

Bookmark

Authors

JHJiacheng Hu XWXM. Wang JLJiawen Lai

Discussion

Member takes

Overview

Systematic evaluation benchmarks accuracy of cardinality estimation methods for data warehousing, suggesting optimal approaches.

Key Points

This paper aims to evaluate the effectiveness of learned cardinality estimation methods on various data warehouse schemas.
Empirical evaluation of seven cardinality estimation methods including query-driven, data-driven, and hybrid approaches.
Benchmarking against PostgreSQL histogram-based estimator across three datasets: TPC-DS, STATS-CEB, and IMDB/JOB.
Measurement of estimation accuracy using Q-Error, inference latency, training cost, and end-to-end query execution time.
Hybrid methods, particularly FactorJoin, achieved the strongest accuracy with a median Q-Error of 1.74 on TPC-DS.
Data-driven methods FLAT and BayesCard achieved a favorable balance between accuracy and inference speed.
BayesCard and FactorJoin maintained high resilience, with a median Q-Error increase of fewer than 1.5 points after a 50% data insertion.

Cite This Study

Hu et al. (2026) studied this question.

synapsesocial.com/papers/6a0d4f4cf03e14405aa9a935 https://doi.org/10.70393/6a6374616d.343134

View Full Paper

Ask AI

Bookmark

May 20, 2026Journal of Computer Technology and Applied MathematicsOpen Access

Benchmarking Learned Cardinality Estimation Techniques for Analytical Query Processing in Data Warehouses

View Full Paper

Ask AI

Bookmark

Authors

JHJiacheng Hu XWXM. Wang JLJiawen Lai

Discussion

Member takes

Overview

Systematic evaluation benchmarks accuracy of cardinality estimation methods for data warehousing, suggesting optimal approaches.

Key Points

This paper aims to evaluate the effectiveness of learned cardinality estimation methods on various data warehouse schemas.
Empirical evaluation of seven cardinality estimation methods including query-driven, data-driven, and hybrid approaches.
Benchmarking against PostgreSQL histogram-based estimator across three datasets: TPC-DS, STATS-CEB, and IMDB/JOB.
Measurement of estimation accuracy using Q-Error, inference latency, training cost, and end-to-end query execution time.
Hybrid methods, particularly FactorJoin, achieved the strongest accuracy with a median Q-Error of 1.74 on TPC-DS.
Data-driven methods FLAT and BayesCard achieved a favorable balance between accuracy and inference speed.
BayesCard and FactorJoin maintained high resilience, with a median Q-Error increase of fewer than 1.5 points after a 50% data insertion.

Cite This Study

Hu et al. (2026) studied this question.

synapsesocial.com/papers/6a0d4f4cf03e14405aa9a935 https://doi.org/10.70393/6a6374616d.343134

View Full Paper

Ask AI

Bookmark