config.py

#
# Configuration parameters
#

# Flag for toggling debugging print statement on or off
debug_output = True

input_dir = './inputs'
output_dir = './outputs'

# input data directly from Itinerum
input_coordinates_file = input_dir + '/coordinates.csv'
input_survey_responses_file = input_dir + '/survey_responses.csv'

# manual ground truth data for comparison with compare.py
locations_gt = output_dir + '/locations_ground_truth.csv'
activities_gt = output_dir + '/episodes_ground_truth.csv'

# timezone to use (until efficient timezone lookups can be implemented)
local_timezone = 'America/Toronto'

# How much time must be spent in one spot for it to be detected as
# an activity episode?
minimum_activity_time = 10*60  # seconds

# Spatial kernel bandwidth in meters (standard deviation of gaussian kernel)
# Used for location detection.
kernel_bandwidth = 25

# Limit of h_accuracy beyond which points get discarded
# (is the unit  meters of in standard deviation of a normal distribution?)
min_accuracy = 100

# Minimum distance between separate locations
location_distance = 100  # meters

# spatial interpolation distance parameter (meters). This is the maximum length
# of a segment to remain uninterpolated for spatial interpolations.
# For reasonable results, this must be < location_distance
interpolation_distance = 30
assert location_distance > interpolation_distance

# Flag for toggling different denominators in calculating episode detection
# Metrics. If on, the denominator consists of the total time for that user,
# If off, it consists of only the portions of episodes that could have affected
# that particular metric, in particular:
# correctly identified unknown time and activity time (not location dependent)
# and misidentified unknown and activity time.
percent_total = True

# Number of worker processes on which to run main.py
# and a flag toggling whether or not to use multiprocessing for main.py
from os import cpu_count
num_pro = cpu_count()
multi_process = False