Machine Learning for Official Statistics and SDG Indicators

This course introduces machine learning as a tool for using either traditional (surveys, micro data,…) or non-traditional data sources (Big Data) to produce high quality predictions for Official Statistics or Sustainable Development Goals (SDGs) indicators.

The course provides an opportunity for participants to explore and comprehend the techniques of machine learning and their links with traditional statistical methods.  It aims at providing an overview of the current methods and applications of Machine Learning, through simplified theoretical concepts, pedagogical case studies and interactive resources. 

The course is not based, nor does it require, a particular software.  However, reproducible examples on either simulated or real data are provided using the R/RStudio environment. Some Python procedures and packages are also provided.   The course has been developed as an interactive training composed of 6 + 1 modules. Each module is composed of several mandatory pedagogical activities, following a logical structure. Activities include videos, interactive videos,  interactive web-based apps, chats, live lectures and webinars, document reading, exercises, polls and quizzes. A preliminary module, M0,  serves as a reminder and  proposes a summary of statistical notions, terminology and basic concepts used by data analysts. These notions will be used throughout the course. 

The course is hosted on the SIAP’s e-learning platform (LMS) which contains a forum for general questions and interactions with the SIAP’s lecturers and e-learning platform administrators. Mandatory weekly webinars, in the form of presentations or Q&A sessions will be proposed using the Microsoft Teams platform. The participants are expected to interact during live sessions and through forums embedded in each module.