SQL-Cardinality-Estimation

An accurate cardinality estimation using Recursive Neural Networks

Cardinality estimation is a key component in query optimization. To choose the best executing plan, the query optimizer should precisely estimate the selectivity of a SQL predicate -- the number of rows in relation selected by the predicate without actual execution.

Getting started

To train a model run following

python3 -m benchmark.models.treernn.train

To reproduce paper results look at this

python3 -m benchmark.evaluation.eval

Reference

If you found this code useful, please cite the following paper:

paper