Hi. As I said before, we can construct a multidimensional model from an OLTP system. Let's see how. There are seven steps: first, define dimensions or the ways to analyze information. Second, identify the attributes of each dimension. Third, organize attributes hierarchically within that dimension. Fourth, assign unique identifiers in the attributes. Fifth, assign description fields in attributes. Sixth, identify the facts and create a facts table. Seventh, assign unique identifiers in the facts table. Given the following entity relationship diagram of 10 multidimensional model: step one, dimensions are logical abstract groupings, such as location, product, and time. Note that time is also a dimension where date is an attribute. Step two, existing entities are grouped into dimensions. An entity becomes an attribute of a dimension. Step three, entities that do not belong to a dimension or that contain facts become fact tables, such as sales and inventory. In this case, city and state are directly related to store because that is the way you want the navigation of the analysis. Note by hierarchy region, state, city, store. Step four, all existing attributes in the original model will be placed in a respective dimension. Step five, by organizing the attributes in dimensions, you get a hierarchy that represents the form of business analysis. They are analysis roots. Step six, the unique identifier must be identified for all the facts. The facts are almost always identified by components, where each component is a key of each attribute relevant to the fact. When necessary, we could include all the unique identifiers for each attributing model. Attributes will either be identified by a single column key or a multi-value key. In the example, all attributes are defined only by a column except for department, which contains a key composed of division ID and department ID. Now, we have finished our first multidimensional model. Congratulations! There are two types of schemas for a multidimensional model, a normalized model or a snowflake, and a denormalized model or star scheme. On the one hand, the snowflake schema contains tables with columns related only to the entity of the real work that represents as its primary key. Every table has a relation, one-to-many to the next table on the same hierarchy forming a dimension. On the other hand, the star schema contains less tables, but each table might contain columns that would not correspond to the same entity of the real work. Every table has a relation one-to-many to the next table on the same hierarchy forming a dimension. Once the multidimensional model has been designed, we can create a data warehouse on the database manager. Remember, that we need to check from which data sources we will feed the data warehouse. After that, we will extract, transform, and load data into the warehouse. Then, analyze the visualized data. Next session, we will understand how to extract, transform, and load data to the warehouse, and how to analyze data with the lab queries by SQL programming.