4.1.2. Problem Simplification

In a rigorous way to evaluate the context impacts, we should depict the performance function *f*(*x*1, *x*2, ... , *xi*, ... , *xn*), 1 ≤ *i* ≤ *n*, where *xi* is the *i*-th context variable such as placement, orientation, sampling rate, subject, algorithm, age, gender, place of experiment, time of the experiment, sensor type and walking velocity. However, it is impossible to collect data under all these context variables. Instead, we should simplify the experiment by restricting some conditions:



**Table 3.** The context variables in the experiments.

In order to evaluate the context impacts of these factors, we should collect data that traverse all the possible cases. Therefore, by exploiting those preliminaries, we conduct our experiments as follows.
