Variable Selection with OxMetrics

Thanks to recent technological developments, researchers, practitioners, and policymakers have now access to very large datasets, also called “Big Data”. In particular cases, users can now even have more variables than observations. For example, it is typically the case when using data from social media. However, in this context of high-dimensional problem traditional statistical and econometric techniques lead to inconsistent result when applied to such large datasets.

The objective of this course is to introduce quantitative methods allowing to reduce information in order to handle high-dimensional problems. Based on classical economic methods (Ordinary Least Squares, Maximum Likelihood Estimator) or principal components, these methods allow automatic selection of variables in high-dimensional problems. The ultimate objective is to study these approaches and to apply them to real data using OxMetrics.

Please note that this is a paid course booked through Timberlake

The sessions are expected to run between 10:00-12:00 and 14:00-16:00. Further details will be confirmed close to the event.