Ravi et al. extracted features from raw data using a window size of 256 samples with 128 overlapping between consecutive windows. Sampling rate used is 50Hz which equivalent to 5.12s for each window. The calculated features are mean, standard deviation, energy and correlation. A single triaxial accelerometer was worn near the pelvic region in the study [1].
Bao and Intille used windowing technique, size of 512 with 256 overlapping between consecutive windows. Sampling rate used is 76.25Hz which equivalent to 6.7s. 5 biaxial accelerometers are used in the study. The calculated features are mean, correlation and energy [2].
[1] N. Ravi, N. D, P. Mysore, and M. L. Littman, “Activity recognition from accelerometer data,” in In Proceedings of the Seventeenth Conference on Innovative Applications of Artificial Intelligence(IAAI, 2005, pp. 1541–1546.
[2] L. Bao and S. S. Intille, “Activity Recognition from User-Annotated Acceleration Data,” in Pervasive Computing, A. Ferscha and F. Mattern, Eds. Springer Berlin Heidelberg, 2004, pp. 1–17.