Unsupervised Learning: Factor and Cluster Analysis in Practice

About

Regression is one of the most commonly used methods for quantitative analyses: it can be applied in almost any field for research purposes, for driving informed decisions or, more generally, for describing our society. Regression analysis is so widely and frequently used, that sometimes one risks to apply it without taking into account important assumptions and prerequisites (for example about the used data). This could lead to misleading, confounding or even “wrong” conclusions (and, consequently, to “bad” decisions).
Thus, it is essential to deeply understand and carefully apply regression analysis methods. In order to reach this objective, we should be aware, first, about how this technique works. In addition, we should be able to explore the available data (and their quality), for making the best decisions about the models we aim at estimating. This knowledge would also enable us to assess the risks we could meet, to evaluate the limits of our results and, finally, to understand whether these findings are reliable.
This course starts from a very simple basis, i.e. from the “raw material” we have at our disposal: data. From this point, step by step, with a simple but complete and rigorous approach, it explains how to better set and develop a regression analysis. Attendees will be able to fully interpret the regression analysis results, evaluating their quality and understanding how “solid” their conclusions are. They will also be able to evaluate and compare potential alternative models and approaches, identifying the best one(s).
Every lecture is structured as a triple-step approach. First, a practical theme is proposed, starting from an example (usually based on a dataset and/or linked to a practical problem or decision to be made). Then, starting from this practical framework, an introduction of the statistical methodology takes place. This is done in order to show how, practically, one could face the issues linked to regression analysis. Finally, a practical part takes place, with the support of Stata. The analysis is set and results are read, interpreted and evaluated, in order to provide an answer to the starting problem.
Thus, the approach starts and is developed from a practical perspective. However, a complete technical introduction (despite in simple terms) will also be made. All lectures will be organized trying to maintain a continuous interaction between the teacher and the attendees.

Course leader

Daniele Toninelli - Università degli Studi di Bergamo

Target group

Professionals, researchers and students who wish to develop high-quality surveys and employ up-to-date statistical methods for survey and quantitative data analyses. Potential attendees include survey practitioners, marketing professionals, social science students, and researchers.

Fee info

Fee 700 EUR, for professionals