Advances in Business Statistics, Methods and Data Collection

23Machine Learning in German Official Statistics¹

Florian Dumpert

Federal Statistical Office of Germany, Wiesbaden, Germany

23.1 Introduction

While machine learning has been tried and often established in many areas of science and economics for some time, official statistics has only begun to address this issue in the last decade. The origins of machine learning go back well into the twentieth century, see for instance the Dartmouth Summer Research Project on Artificial Intelligence (Samuel 1959), or the theoretical anticipation of the methods, etc. However, many of this field's approaches only gained some applicability outside of research and in official statistics as computing power and large (not necessarily official) data sets became increasingly available and legally accessible.

This chapter aims to provide an overview of the use of machine learning in official statistics, in particular in Germany. However, this goal is not to be achieved by way of a complete list of projects. Rather, the necessary strong link between science and research, national and international exchange and collaboration and specific application shall be revealed and explained by way of examples. The examples are taken from the field of earnings statistics, all of whose data are collected from reporting establishments in the form of samples.

The chapter presents in the second section a short introduction to machine learning, in Section 23.3 an overview of machine learning in official statistics in ...