Setting parameter values is crucial for the performance of metaheuristics. Tuning the parameters of a metaheuristic is a computationally costly task. Moreover, parameter tuning is difficult considering their inherent stochasticity and problem instance dependence. In this work, we explore the application of machine learning algorithms to suggest suitable parameter values. We propose a methodology to use k-nearest neighbours and artificial neural network algorithms to predict suitable parameter values based on instance features. Here, we evaluate our proposal on the Capacitated Vehicle Routing Problem with Time Windows (CVRPTW) using its state-of-the-art algorithm, Hybrid Genetic Search (HGS). Additionally, we use the well-known tuning algorithm ParamILS to obtain suitable parameter configurations for HGS. We use a well-known instance set that considers between 200 and 1000 clients. Three sets of features based on geographical distribution, time windows, and client clustering are obtained. An in-depth exploratory analysis of the clustering features is also presented. The results are promising, demonstrating that the proposed method can successfully predict suitable parameter configurations for unseen instances and suggest configurations that perform better than baseline configurations. Furthermore, we present an explainability analysis to detect which features are more relevant for the prediction of suitable parameter values.
University students are constantly exposed to high-tension situations and peaks of stress caused by the difficulty from their many responsibilities. These situations can produce disorders in students, such as psychosomatic, behavioral, or emotional disorders. Burnout is a work-related syndrome that can be observed in students. This syndrome considers depersonalization, emotional exhaustion, and diminished feelings of personal accomplishment. We aim to detect anxiety, depression, and burnout symptoms in university students to prevent further negative consequences. For this, we design a questionnaire using well-known instruments to detect these signs. We propose to use an ensemble of classifiers, including random forest and artificial neural networks, to predict a set of four possible disturbances in persons. The proposal will be used in the Human Place project to suggest strategies to tackle four disturbance types. This study considers the participation of 93 persons from the Valparaiso region in Chile. Results show that the evaluated algorithms can predict the presence or absence of the disturbances with high accuracy levels and a low number of false negative cases. We also present a detailed analysis of which questions were relevant in the classification task of each algorithm.