The physical monitorization using dedicated devices is becoming an everyday routine for an increasing number of people. The information
provided by accelerometers enables the creation of a diary of the
performed activities, and the determination of their intensity. The aim
of this study is to evaluate the potentiality of the smartphone's
accelerometer to perform such an activity. We developed an application
to capture the signal from the smartphone's accelerometer, when it is
positioned along the waist in the front pocket of an individual, in an
attempt to create the most natural conditions possible. The study
explored features extracted in both time and frequency domain, and
parametric and non-parametric classifiers. Preliminary results demonstrate that the classification of activities can be done with remarkable accuracy ($>$ 95\%)