Business Analytics and OLAP

Business Analytics

• Business Analytics (BA) is a technique used to visualize and analyze business data, which aids in decision-making.
• Analytics refers to the science of analysis.

OLAP (On-Line Analytics Processing)

• OLAP stands for On-Line Analytics Processing.
• OLAP offers fast analysis of shared multidimensional information
• The three core features of OLAP are that it is fast, focuses on analysis, and is multidimensional.
• Provides quicker response times than OLTP, around 0.1%.
• Designed for querying and generating reports.
• OLAP systems are based on cubes.

OLAP vs OLTP (On-Line Transactional Processing)

• OLAP is analytical whereas OLTP is transactional.
• OLTP focuses on processing repetitive transactions in large volumes through simple manipulations.
• OLAP focuses on examining complex relationships among numerous data items.
• OLAP can analyze relationships, identifying patterns, trends, and exceptions.
• OLAP serves as a direct decision support method.
• OLTP is difficult to query.
• OLAP structures are simplified for use by the analysts and business users.
• Simplification is possible because OLAP is read-only.
• OLAP systems require significantly more storage space than OLTP.

OLAP Database Size

• Big companies often have vast amounts of data.
• OLAP database sizes are measured in Terabytes or Petabytes instead of Gigabytes.
• Albert Heijn is a supermarket chain example that has 800 stores in the Netherlands.
• Each store has approximately 15 products and an average checkout time of 1 minute.
• Each store averages about 5 checkouts per day.
• A single store will generate 54,000 records per day.
• All 800 Albert Hejin stores will generate 43,200,000 records per day.
• Over the course of a year, the records grow to 15,379,200,000.
• YouTube has approximately 65,000 videos uploaded per day.
• Approximately 1,950,000 videos are uploaded each month.
• If each video is 1MB, YouTube grows by 1.86 terabytes each month.
• If each video is 10MB, YouTube would grow by 18.6 terabytes each month.
• 1 Terabyte (TB) is equal to 1430 CDs.
• 1 Petabyte (PB) in piled-up CDs would be 1.8 km high.

Cubes

• Cubes are based on dimensions, measures and hierarchies.
• Multidimensional refers to considering data in multiple dimensions for analysis.
• Measures are fact table, which represents the data being analyzed.
• Hierarchies are dimensions used to organize and navigate the data.

Slicing

• Slicing fixes one dimension at a specific value.
• When slicing, a slice allows selecting a subset of the cube's data by limiting one of the cube's dimensions to data that meets a specific condition.
• For example, finding all sales numbers only for Lithuania, Europe, or Vilnius.

Dicing

• Dicing fixes several dimensions at specific values resulting in a subcube.
• Dicing allows for defining a subcube for which selection criteria for two or more dimension can be specified.
• For example, viewing all sales numbers only for Lithuania and the Netherlands, AND only for internet and mail order sales.

Drill Down

• Drill-down allows focusing closely into the details of the data.
• It uses hierarchies to move e.g. from Europe to Lithuania and finally to Vilnius.
• Drill-down enables navigation from generalized to more detailed data.
• Is a repetitive process of queries; for example, sales profit is analyzed first by year, then by quarter, and finally by month.

Roll Up

• Roll-up is the logical opposite of the drill-down.
• It involves zooming out from the data.
• For example, going from Vilnius to Lithuania to Europe.
• Aggregates data using a given hierarchy.

Pivot Tables for Data Presentation

• Display information generated by OLAP.
• Tables can change the columns of a table.
• Pivoting (rotating) enables rotation of the axes to visualize data.