A correlation database is an unconventional, date-model-independent DBMS (database management system) that productively handles unplanned queries. Unlike the relational database management approach which considers rows and columns as a data storage approach and indexes a set of data in a whole row, a correlation database, on the other hand, utilizes a value-based storage approach where each data entry (value) is given an indexing id.
The correlation database stores each new value once and gives it a numeric identifier. It does not matter how many times the value is occurring and what's its location in the original dataset. Then the original data is modified by linking to those numeric identifiers. Correlation database does data normalization as it stores it in the dataset.
Correlation can be done manually or a person can use tools like Microsoft Excel. Three possible types of correlation are positive correlation, negative correlation, and no correlation.