Empresas comerciales zombis ecuatorianas: Una clasificación mediante Machine Learning

In this article, we have studied which Machine Learning methodology most accurately predicts the presence of zombie companies in the Ecuadorian commercial sector. To do this, data from these companies in the years 2019, 2020 and 2021 were used. The zombie variable was defined as a binary variable that took the value of one if the company had had negative equity in the previous three years, and zero in another. case. The results determined that the different Machine Learning methods are accurate when predicting zombie companies, although logistic regression yields the best results in terms of ROC curves in the years 2019 and 2020, with the rest of the methods not being very distant in terms of results. For the rest of the metrics, Random Forest is the best independent technique of the year studied.