Power BI
Intro to BI
- Business Intelligence or BI is using data to make better decisions
- Better decisions to take the business forward
- Examples:
- ways to attract new or retain old customers
- competitor analysis
- what is driving profits?
- what expenses can be diminished?
- Key concepts:
- Domain
- The context in which BI is applied
- Ex. std business functions or depts like sales, marketing etc.
- Helps narrow focus
- Data
- Next step after selecting a domain is selection of relevant data
- That means finding sources of relevant data:
- Internal Data
- Data generated within the boundaries of an org
- net revenues, units produced, sales etc.
- External Data
- BI is most effective when internal data is combined with external data
- data created outside the boundaries of an org
- Eg. global economic performance, census info, competitor prices etc.
- Also types of data:
- Structured
- Data that conforms to a specific structure with rows and columns
- Mostly relational database systems and database standards (ODBC, OLE DB) fall in this category
- Unstructured
- Data that cannot be organized into standard rows and columns
- Videos, audio, images, text
- word, pdfs, emails, social media posts
- these are most difficult to consume and analyze
- stored as BLOBS (binary large objects) or as a file in file systems (NTFS, HDFS)
- No-SQL databases also fall under this category
- Semi-structured
- Structured data but not conforming to row/column standard
- Delimited text files, XML, markup languages, JSON, EDI
- They have self-defining structure which makes them easy to consume and analyze but harder than true structured data
- data access protocols like OData / REST also fall in this category
- BI tools are optimized for handling structured and semi-structured data
- BI tools are designed to ingest semi-structured data sources and transform them to structured
- Model
- A data model refers to the way in which one or more data sources are organized
- The organization is done to support analysis and visualization
- Key steps:
- Organizing
- Organzing is done by establishing how the multiple data sources relate to one another
- This helps model become a cohesive whole
- For example the relation between a CRM data and ERP data can be done based on customer name
- Transforming and cleansing
- It is almost always necessary to clean the source data
- Duplicates, trailing spaces, spelling mistakes, missing data
- Transforming and cleansing tech are often called ETL and have special tools
- Defining and cetegorizing
- Defining data types on the source data eg. date, time, datetime etc.
- Defining catergories in each data type for example a text can be a url
- These provide ways to establish what analysis can be done on each type
- Analysis
- Grouping data, simple aggregations, sums, counts and averages
- sPecialized calculations to identify trends, correlations and forecasting
- Creating KPIs
- There are tons of tools to do the analysis. R, Python and SQL are top ones along with Excel.
- Visualization
- Visualizations are tables, matrices, pie charts, bar graphs, and other visual displays that help provide context and meaning to the analysis
- Business intelligence tools allow multiple individual tables and charts to be combined on a single page or report.
Power BI ecosystem
- Power BI is a collection of interralted tools and services that form a complete BI ecosystem.
- It includes tools for modelling, analysis and visualization
- The ecosystem can be broken down into following categories:
- Core - Power BI specific
- Power BI Desktop:
- Free, Windows based app that installs on a local computer
- Its primary tool to ingest, cleanse, transform data, combine into models
- Analyze and visualize using calculations, visualizations, and reports
- Once reports are created in Power BI desktop they are published to Power BI service
- Power BI service:
- Cloud-based SaaS online platform
- light report editing, sharing, collaborating, and viewing reports
- Core - non Power BI specific
- Data Query - Data connectivity and transformation
- DAX - programming language for Power BI
- On Premise data gateway - facilitate access from Power BI service to data sources
- SSAS - Ability to build models
- MS Appsource - Marketplace to find apps, add-ins and extensions to sotwares like Power BI
- Non-core - Power BI specific
- Report Server
- Embedded
- Mobile application
- Mixed reality
- Natively integrated MS tech
- Office 365, Excel, Microsoft Flow, Power Apps, Visio, Azure ML, Report builder
- Extended ecosystem
- Large ecosystem of 3rd party tools and add-ons
graph LR
A[Power BI] --> B[Power BI Desktop];
A --> C[Power BI Services];
B --> D[Getting Data];
B --> E[Creating a data model];
B --> F[Analyzing Data];
B --> G[Creating & Publishing reports];
C --> H[View and Edit reports];
C --> I[Creating Dashboards];
C --> J[Sharing and collaborating];
C --> K[Accessing and creating apps];
C --> L[Refresh data]
Power BI Desktop
Connecting and Shaping Data
Data Models and Calculations
Unlocking Insights
Final Report
Publishing and Sharing
Using reports in the service
Dashboard, Apps & security
Gateways and refreshing Datasets