Predictive Analytics

Abstract Although humans have long sought to produce the most accurate predictions possible about the future and apply them to make decisions in the present, recent developments in machine learning have led to predictions assuming an even more important role in society. There is hardly any societal field in which algorithmic forecasts are not carried out and hardly anyone who has not become the subject of forecasts in their life. Against this backdrop, this chapter presents the existing literature on the topic and reflects on how predictive analytics should be understood from a sociological perspective and on the consequences with which it may be associated. It will show that predictive analytics needs to be grasped as a socio-technical constellation because any genuine reference to the future does not result from the technical-analytical process itself but comes about only through human interpretation of the results as future oriented. In addition, the chapter discusses the human role in the collection of data, in the process of algorithmic model construction, and in the selection and creation of patterns. It also highlights the fact that predictions cannot be separated analytically from their final implementation, as this has important consequences that often follow a circular logic and can, ultimately, have discriminatory consequences.