Browsing by Subject "cumulative distribution function"

Indexes are data structures that are used for retrieving records from a database. They are used in database management systems (DBMS) to optimize queries. The abundance of available data has motivated the research of indexes to provide faster query times and to have smaller memory usage. With the advantages of machine learning, new variations of indexes have been created. These indexes utilize the data distribution for faster query times and smaller footprints in comparison to traditional indexes like B-tree and B+ tree. These indexes are known as the learned indexes. In this thesis, we study the effect the distribution change between normal distributions has on one-dimensional learned indexes and compare that to the B+ tree. We conduct an experiment where we simulate the distribution change and measure the insertion and query times. In this experiment, we include three learned indexes, which are ALEX, PGM, and LIPP.