Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/sarathchandranpm/data-modelling-creating-a-star-schema

Build a star schema by splitting the original dataset into facts and dimension tables
https://github.com/sarathchandranpm/data-modelling-creating-a-star-schema

Last synced: about 2 months ago
JSON representation

Build a star schema by splitting the original dataset into facts and dimension tables

Awesome Lists containing this project

README 

        
# Sales Data Model Project in Power BI - Part 1



## Overview

This part of the project demonstrates the implementation of a dimensional data model (primarily star schema with snowflake elements) in Power BI using a sales dataset. The model transforms a flat CSV file into an optimized dimensional structure, improving data analysis capabilities and reporting performance.



## Data Source

The original dataset was provided as an Excel CSV file containing sales transactions with the following fields:

* ProductID, Date, CustomerID, CampaignID, Units, Product, Category, Segment

* ManufacturerID, Manufacturer, Unit Cost, Unit Price

* Customer details: ZipCode, Email, Name

* Geographic information: City, State, Region, District, Country



## Data Model Architecture



![image](https://github.com/user-attachments/assets/54a66e65-240d-4a05-923b-e332c2ddb0b4)



### Fact Table

The central fact table **Sales** contains the core transaction metrics and foreign keys:

* Measures: Units, Unit Cost, Unit Price

* Foreign Keys: ProductID, CustomerID, Date

* Additional Fields: Date, Manufacturer



### Dimension Tables

The model includes four dimension tables:



#### 1. Dim Products

Contains product-related attributes:

* ProductID (Primary Key)

* Product

* Category

* Segment



#### 2. Dim Customers

Stores customer information:

* CustomerID (Primary Key)

* Email

* First Name

* Last Name

* ZipCode (Foreign Key to Dim Geography)



#### 3. Dim Geography

Contains location hierarchies:

* ZipCode (Primary Key)

* City

* State

* Region

* District

* Country



*Note: This dimension introduces a snowflake element as it connects to the fact table through Dim Customers rather than directly.*



#### 4. Dim Date

A dedicated date dimension marked as the official date table:

* Date

* Day of Week

* Month

* Month Number

* Quarter

* Week Number

* Year



## Model Characteristics



### Star Schema Implementation

The model primarily follows a star schema design with these key characteristics:

* Centralized fact table (Sales)

* Denormalized dimension tables

* Single join path between fact and most dimensions

* Simplified query paths for better performance



### Snowflake Element

The geographic dimension (Dim Geography) introduces a snowflake element by:

* Connecting to the fact table through Dim Customers

* Using ZipCode as the linking field between Dim Customers and Dim Geography



## Development Process

The project followed a structured approach to data modeling:



1. **Conceptual Modeling**

   * Identified core business entities

   * Defined relationships between entities

   * Determined fact and dimension candidates



2. **Logical Modeling**

   * Designed the star schema structure

   * Defined primary and foreign keys

   * Established table relationships



3. **Physical Modeling**

   * Implemented the model in Power BI

   * Created appropriate data types

   * Established relationships with correct cardinality

   * Optimized for performance



## Learning Outcomes

This project provided valuable insights into:

* Star schema design principles and implementation

* Data normalization importance and techniques

* Dimensional modeling best practices

* Power BI data modeling capabilities

* The balance between performance and model complexity



# Sales and Budget Data Model Enhancement - Part 2



## Overview: 

This phase of the project extends the existing sales data model by incorporating budget data, demonstrating advanced data modeling techniques to handle many-to-many relationships and complex dimensional hierarchies.



## New Data Structure



### New Fact Table

**FactBudget** was introduced with the following fields:

* Value (measure)

* Date (foreign key)

* Scenario

* Original Category and Segment fields (later replaced with CatSeg Key)



## Data Modeling Challenges and Solutions



### Handling Many-to-Many Relationships

* Identified a potential many-to-many relationship between FactBudget and product dimension tables through Category

* Implemented a bridge table solution to resolve the many-to-many relationship

* Created a new dimension table (Dim CatSeg) to establish proper relationships



### Creation of Bridge Dimension

**Dim CatSeg** was created through the following steps:

* Extracted Category and Segment combinations from Dim Category

* Removed duplicates to ensure distinct Category-Segment pairs

* Created a surrogate key (CatSeg Key) using the index column technique

* Established as the central dimension for category-related attributes



### Data Transformation Process

1. **Merge Operations**

   * Merged Dim CatSeg with FactBudget to incorporate CatSeg Key

   * Performed similar merge operation with Dim Product

   * Based merge operations on Category and Segment columns



2. **Data Cleanup**

   * Removed redundant Category and Segment columns from FactBudget

   * Cleaned up Dim Product by removing duplicate category information

   * Maintained data integrity while reducing model complexity



## Model Optimization



### Relationship Management

* Established proper relationships between:

   * FactBudget and Dim Date through Date column

   * FactBudget and Dim CatSeg through CatSeg Key

   * Dim Product and Dim CatSeg through CatSeg Key

     

## Data Model Architecture

**Final Data Model:**

![image](https://github.com/user-attachments/assets/7d8e4451-9fc1-4686-94c8-4de4c2ab08f7)

**Sales table and related dimension tables:**

![image](https://github.com/user-attachments/assets/5a0eb62f-3f1b-4b74-8d38-7d1fdfe4fc07)

**FactBudget table and related dimension tables:**

![image](https://github.com/user-attachments/assets/5d8c4a3d-dcc5-43ed-b2c4-995d556be7dd)



### Performance Considerations

* Implemented surrogate keys for efficient joins

* Removed redundant columns to reduce model size

* Optimized relationship cardinality for better query performance



## Technical Skills Demonstrated



### Advanced Data Modeling

* Resolution of many-to-many relationships

* Creation and implementation of bridge tables

* Surrogate key generation and management



### Data Transformation

* Complex merge operations

* Duplicate removal and data deduplication

* Column management and cleanup



### Dimensional Modeling

* Bridge dimension design and implementation

* Proper key management

* Relationship optimization



### Power BI Features Utilized

* Merge queries functionality

* Index column creation

* Relationship management

* Data model optimization techniques



## Business Intelligence Best Practices

* Maintained dimensional modeling standards

* Implemented proper key management

* Ensured data integrity across the model

* Created efficient paths for data analysis



## Results

The enhanced data model now supports:

* Integrated analysis of sales and budget data

* Proper handling of category-segment hierarchies

* Efficient querying across multiple fact tables

* Streamlined reporting capabilities



This phase of the project demonstrates advanced data modeling techniques and showcases the ability to handle complex business requirements while maintaining model efficiency and analytical capabilities.



## Tools Used

* Power BI Desktop

* Excel (source data)