Towards the Recommendation of Time for Physical Activities Based on Air Pollution and Meteorological Variables

Exercising outdoors, in a polluted environment, can cause adverse health effects for people. Therefore, it is important to know the levels of pollutants in the environment in which the exercise is carried out. This article applies the Clustering technique to generate a recommendation system of hours of the day in which it is possible to perform physical activities, reducing the damage to health, considering the levels of pollutants present in the environment. A dataset provided by the Monitoring Network of the Public Mobility, Transit and Transport Company (EMOV EP) of Cuenca, Ecuador, was used. The results show that through an unsupervised learning data mining technique such as clustering, a recommendation system can be implemented. This system generates a range of time within physical activities are suggested to be performed, reducing the negative impact on people’s health of high levels of pollutants and meteorological variables present in the environment.