Anàlisi causal: estudi de les relacions causa-efecte i aplicació a l’anàlisi de dades

Abstract

[en] Causal inference is a branch of mathematical analysis that has been poorly studied until recently. The difficulties involved its research and the high number of personal decisions the researcher has to make in these types of studies have often moved it away from the statistical side, although much of the mathematical basis is shared. In this work, we review the formal definition of causality, analyze the problems involved in the lack of complete information from the two worlds we observe (with treatment and without), explain the theory that allows us to eliminate some of these problems and seek practical methodologies for causal study both of observational and non-observatory experiments. Finally, we have a practice with the analysis of the impact of adverse weather conditions (moderate and intense rainfall) on the use of bicycles in the city of Barcelona. We conclude that moderate rain reduces the use of the bicycle by 40% on average, although it varies dependin on the month, with months when the impact is almost zero and others reaching 70%.