Analytics Overview and Phases

Study Notes

Analytics uses arithmetic, statistical, data mining, and advanced methods like machine learning to find new parameters and information that add value to data.
It enables building models based on selecting the right data, which are then tested and used for services and processes.

Analytics has three phases: descriptive, predictive, and prescriptive analytics.
These phases derive new facts and provide business intelligence.

Descriptive analytics answers questions about what happened in the past.
It involves finding aggregates, frequencies, mean values, and variances in values or groupings using selected properties.
It enables:
- Online Analytical Processing (OLAP) for analytics
- Reporting or generating spreadsheets
- Visualisations or dashboard displays of the analysed results
- Creating indicators, called key performance indicators

Spreadsheet-based reports and data visualisations:
- Present results in a spreadsheet format
- Enable user visualisation of "what if" scenarios
- Use tables with values in cells that can be related to other values through formulas or statistical analysis
Descriptive statistics-based reports and data visualisations:
- Use statistical analysis to find peaks, minima, variance, probabilities, and statistical parameters
- Use formulae to enable data showing variations
Data mining and machine learning methods in analytics:
- Use algorithms to extract hidden or unknown information from large amounts of data
- Model specific tasks using machine learning
Online analytical processing (OLAP) in analytics:
- Enables viewing of analysed data up to the desired granularity
- Enables rollup or drill down of data
- Provides summarized information and automated reports from large volume databases

Uses algorithms, such as regression analysis, correlation, optimisation, and multivariate statistics
Uses techniques, such as modelling, simulation, machine learning, and neural networks
Enables:
- Predicting trends
- Undertaking preventive maintenance
- Managing campaigns with integrated marketing strategy
- Predicting by identifying patterns, clusters, and anomalous characteristics
Results of predictions need verification from domain knowledge and multiple angles

Answers not only what is anticipated or what will happen but also why it will happen
Suggests actions for deriving benefits from predictions
Shows the implications of decision options or optimal solutions
Provides new resource allocation strategies or risk mitigation strategies
Suggests the best course of actions in a given state or set of inputs and rules

Events are definable options that are unique, non-interaction, or interaction options
Event analytics uses event data for events tracking and event reporting
Enables tracking and reporting of events