The dataset used here is Microsoft Contoso Retail Classic. The data has huge number of transactions and has 20+ tables and 16M+ transactions in fact tables. The Dataset first was downloaded and imported into SQL Server using the SQL backup function. The downloaded table was in normalized form and in Snowflake schema.

​

​

Next, ETL was prformed using SQL queries for the major transformations. During the ETL, two fact Sales tables were merged into one big table and a single primary key SalesKey was added. Columns which were not relevant for the analysis were removed and truncated such that a consistent fact table can be obtained.

​

​

Later the model was passed through Power Query for further transformations where redundant tables & columns were removed. Here, Product Category & Product Subcategoty tables were denormalised by merging them through a common column. The data was next imported in Power BI where Data modeling took place and model was converted to a Star Schemas and various Fact & Dimension were joined forming One-to-Many relationships. Finally, after adding the required DAX measures and Visuals, the analysis was performed.