In the last decades the amount of available data increased extremely and the data analytics field evolved and expanded accordingly. Starting from linear regression many other algorithms have been developed, leading to the spread of machine learning methods. Such methods are able to implement more complex regression models, achieving higher accuracy, but often at the cost of interpretability. In this master's thesis a study about the interpretation of some regression models has been carried out. Methods to evaluate variable importance and to display efficiently feature effects on the variable of interest are explained; the respectively R functions that implement such methods are illustrated through the presentation of a case study developed by the author during an internship in Tetra Pak.