What is behavioural modelling?

Introduction

Behavioural modelling is all about exploring different dimensions (also referred to as behavioural features) and identifying the ones which are most closely related to a particular objective as represented by your analysis selection. If, for example, the analysis selection is people who respond to a particular campaign, behavioural modelling uses historic transactional data to identify indicative behaviours of responders (for example, they have high website engagement in the three months prior to the campaign) so that a model can be created to then predict which other customers might respond based on their current behaviour.

Standard modelling

FastStats models have always been able to predict likely purchasers of a product based on the characteristics of the current purchasers. This gives you a useful tool to help focus your marketing, using socio-demographic and some transactional details. For example, if you know who has already been to Sweden, you can use a Decision Tree or Profile to identify their characteristics and find ‘lookalikes’ to whom you can try and sell holidays to Sweden.

Behavioural modelling

Behaviour-based modelling in the FastStats Modelling Environment develops modelling functionality in two very useful ways:

  1. It allows you to use transactional data to a far greater extent, by creating dimensions that can explore transactional patterns and behaviours. For example, if your system has a cost variable, rather than just explore the cost of bookings, you could explore the cost of bookings in a specific period prior to a campaign, or even if people are spending more or less on their holidays as time progresses.

  2. It provides an opportunity to create the model at one point in time, evaluate it at another point in time and then, if the evaluation shows the model is effective, apply the model to the present date.

 

See also: