How can I reduce the caputured movement data of a person in a way, that I have filtered the main features of the movement. Or how can I detect pattern/ main features in that data set?
I captured some data with a device (Inertial sensor and motion capturing (x,y,z)) attached to human body. So I have a huge data set. I want to prepare the data set so I have no noise and no "unecessary" data.
In the end I just want to know for example that certain movement behaviour hints to a clueless guy or a person under stress.
I first thought approaches from association analysis could be useful but I think based on what I have read about the application of such algorithms they are not suitable for this data set.