An overview of the four main approaches to predictive analytics

This infographic provides an overview of the four main families of approaches to predictive analytics.
  • Prediction encompasses applications that aim to estimate or predict the values of a key target field.
  • Segmentation refers to techniques such as cluster analysis which attempt to find the most ‘naturally occurring” groups within a dataset.
  • Association modelling discovers groups of categories that are likely to co-occur such as items in a shopping basket.
  • Forecasting refers to methods that extrapolate time trends such as sales of products so businesses can anticipate future demand.

Each family is further detailed with colour coded examples of the popular methods that underpin the approach. Green indicates methods from statistical approaches, orange denotes techniques that are based on uncovering rules. Finally the techniques coloured blue refer to machine learning algorithms.