Here we'll discuss best practices for analytics tasks

There are four types of data analytics:

1. Descriptive: what is happening? This is the most common form of analytics. It provides the analyst a view of key metrics and measurements within the problem of study.
2. Diagnostic: why is it happening? This step represents a typical data analysis, trying to understand the root-cause of the problem by data exploration phase, e.g. obtaining aggregated information, doing basic statistical analysis, performing data visualization.
3. Predictive: what is likely to happen? This step represents the forecast of a particular problem usually done in the form of modeling events based on historical information. At this step a data analyst usually applies Machine Learning to predict the future events on a given dataset.
4. Prescriptive: what do I need to do? At this step an analyst exploits the complexity and value of the applied model. We try to understand what has happened, why it has happened and what might happen to help determine set of actions to attack the underlying problem. Usually prescriptive analysis is not concentrated on an individual action but rather propose and implement a series of them.