BI (Business Intelligence)

technology, processes and practices used to collect, integrate, analyze and present business information

transforms raw data into useful insights

main objective: support better data-driven decision-making in organizations
- make informed, strategic decisions
- identify business trends and patterns
- improve operational efficiency
- monitor Key Performance Indicators (KPIs)
- gain competitive advantage
components:
- data visualization
  - e.g. power bi (microsoft), tableau, google data studio
- data collection
- data storage
  - e.g. amazon redshift, google bigquery, snowflake
- ETL tools
  - e.g. apache NiFi
- OLAP tools
  - e.g. ibm cognos, microsoft analysis services (SSAS)
- data analysis
- decision support
bi cycle
- data collection
- data processing and transformation
- data storage
- data analysis
- decision-making
BI vs data science vs data analytics
- BI: descriptive (what happened?)
- data analytics: diagnostic (why did it happen?)
- data science: predictive/prescriptive (what will/could happen?)

raw data => tools that extract and transform data => storage tools <= analysis tools

OLTP (OnLine Transaction Processing)

OLTP systems are built to handle real-time transactional data

handles day-to-day transactions
used for transactional processing of data in real-time
used for processing business transactions such as order processing, inventory management, etc
normally uses postgresql, mysql, sql server

OLAP systems are built for analying large volumes of data

handles analysis on large volumes of data and supports decision-making
used for platforms with intense read operations
used for analyzing large quantities of data from multiple perspectives or dimension
normally uses amazon redshift, google bigquery, snowflake

data cube: representation of multidimensional data
- each dimension is an axis
- each cell in the cube contains measured data (aka facts)
- sub-cube: subset of the full data cube
slice: selects single value for one dimension and creates a sub-cube by slicing along that dimension
dice: selects sub-cube by choosing specific values for two or more dimensions
drill down: increase detail + reduce granularity
- e.g. yearly sales => quarterly => monthly => daily
roll up: reduce detail + increase granularity
- e.g. daily sales => monthly => quarterly => yearly
pivot (rotate): reorients the multidimensional view to provide an alternative presentation of data
- e.g. switch rows and columns

KPI (Key Performance Indicator): measures how effectively an organization is achieving key business objectives
data mining: process of automatically discovering useful patterns, trends, correlations or rules in large dataset