Unsupervised Model with Cardinality Estimation

Abstract

In the field of database research, query optimization is an important part. Cardinality estimation is a critical problem in query optimization. Reasonable and accurate cardinality estimation can make it easier for the query optimizer to produce a good connection order and guide the cost estimation of the query optimizer. This paper focuses on using the unsupervised method to deal with the cardinality estimation problem, learning the specific distribution in the data table through unsupervised learning, and producing good cardinality estimation results. At the same time, it also experimented with the distribution of actual data and finally achieved good experimental results.