def infer_nonwear_triaxial(x, y, z, minimum_length=timedelta(hours=1), noise_cutoff_mg=13, return_nonwear_binary=False):
''' Use the 3 channels of triaxial acceleration to infer periods of nonwear '''
# Get an exhaustive list of bouts where the monitor was still
x_intersect_y_intersect_z = infer_still_bouts_triaxial(x,y,z, noise_cutoff_mg=noise_cutoff_mg, minimum_length=minimum_length)
# Restrict those bouts to only those with a length that exceeds the minimum length criterion
x_intersect_y_intersect_z = Bout.limit_to_lengths(x_intersect_y_intersect_z, min_length=minimum_length)
# Legacy code - probably going to delete this
if return_nonwear_binary:
# Create a parallel, binary channel indicating if that time point was in or out of wear
nonwear_binary = Channel.channel_from_bouts(x_intersect_y_intersect_z, x.timeframe, False, "nonwear", skeleton=x)
return (x_intersect_y_intersect_z, nonwear_binary)
else:
return x_intersect_y_intersect_z
def produce_binary_channels(bouts, lengths, skeleton_channel):
Bout.cache_lengths(bouts)
bouts.sort(key=lambda x: x.length, reverse=True)
channels = []
for length in lengths:
# Drop bouts from list if their length is less than x minutes
bouts = Bout.limit_to_lengths(bouts, min_length=length, sorted=True)
channel_name = "{}_mt{}".format(skeleton_channel.name,length)
# Clone the blank channel, set data to 1 where time is inside any of the bouts
skeleton_copy = copy.deepcopy(skeleton_channel)
chan = Channel.channel_from_bouts(bouts, False, False, channel_name, skeleton=skeleton_copy)
channels.append(chan)
return channels
