Pentaho Analysis is built on the Mondrian online analytical processing (OLAP) engine. OLAP relies on a multidimensional data model that, when queried, returns a dataset that resembles a grid. The rows and columns that describe and bring meaning to the data in that grid are dimensions, and the hard numerical values in each cell are the measures or facts. In Pentaho Analyzer, dimensions are shown in yellow and measures are in blue.
OLAP requires a properly prepared data source in the form of a star or snowflake schema that defines a logical multi-dimensional database and maps it to a physical database model. Once you have your initial data structure in place, you must design a descriptive layer for it in the form of a Mondrian schema, which consists of one or more cubes, hierarchies, and members. Only when you have a tested and optimized Mondrian schema is your data prepared on a basic level for end-user tools like Pentaho Analyzer. See Workflow Overview for a more comprehensive overview of the Pentaho Analysis data preparation workflow, including which Pentaho tools you will need to execute this process.
For concise definitions of OLAP terms, refer to Mondrian Schema Element Quick Reference and the individual element pages it references.