“…The features used are mean, variance, standard deviation, median, minimum, maximum, range, Interquartile range, Kurtosis, skewness and spectrum peak position of the accelerometer data, reporting an accuracy of 93.8% [25]. Another authors implemented the Sliding-Window-based Hidden Markov Model (SW-HMM), and compared this method with SVM and ANN for the recognition of walking, standing, running, going up stairs, and going down stairs activities, using the mean, variance and quartiles of the accelerometer data [26], reporting an accuracy around 80%.The J48 decision tree, Random Forest, Instance-based learning (IBk), and rule induction (J-Rip) methods were used with accelerometer data for the recognition of standing, sitting, going up stairs, going down stairs, walking, and jogging, implementing the Dual-tree complex wavelet transform (DT-CWT), DT-CWT statistical information and orientation as features, reporting an accuracy of 86% for the recognition of all activities [27].The authors of [28] created a system named Actitracker, that performs the recognition of walking, jogging, going up stairs, going down stairs, standing, sitting, and lying down activities, using the Random Forest method and accelerometer data. This systems uses the mean and standard deviation for each axis, the bin distribution and the heuristic measure of wave periodicity, with an accuracy around 90% [28].In [10], the authors implemented a solution using ANN and SVM methods applied to the accelerometer data, in order to identify several activities, such as standing, sitting, standing up from a chair, sitting down on a chair, walking, lying, and falling activities.…”