def classify(classifierList):
'''
This function wraps other functions in order to load, classify, and return the label for each 5 second epoch of Q sensor data.
INPUT:
classifierList: list of strings, either "Binary" or "Multiclass"
OUTPUT:
featureLabels: Series, index is a list of timestamps for each 5 seconds, values of -1, 0, or 1 for artifact, questionable, or clean
data: DataFrame, only output if fullFeatureOutput=1, index is a list of timestamps at 8Hz, columns include AccelZ, AccelY, AccelX, Temp, EDA, filtered_eda
'''
# Constants
oneHour = 8 * 60 * 60 # 8(samp/s)*60(s/min)*60(min/hour) = samp/hour
fiveSec = 8 * 5
# Load data
data, _ = getInputLoadFile()
# Get pickle List and featureNames list
featureNameList = [[]] * len(classifierList)
for i in range(len(classifierList)):
featureNames = getSVMFeatures(classifierList[i])
featureNameList[i] = featureNames
# Get the number of data points, hours, and labels
rows = len(data)
num_labels = int(np.ceil(float(rows) / fiveSec))
hours = int(np.ceil(float(rows) / oneHour))
# Initialize labels array
labels = -1 * np.ones((num_labels, len(classifierList)))
for h in range(hours):
# Get a data slice that is at most 1 hour long
start = h * oneHour
end = min((h + 1) * oneHour, rows)
cur_data = data[start:end]
features = createFeatureDF(cur_data)
for i in range(len(classifierList)):
# Get correct feature names for classifier
classifierName = classifierList[i]
featureNames = featureNameList[i]
# Label each 5 second epoch
temp_labels = classifyEpochs(features, featureNames,
classifierName)
labels[(h * 12 * 60):(h * 12 * 60 + temp_labels.shape[0]),
i] = temp_labels
return labels, data
if x_seconds:
plt.xlabel('Time (s)')
else:
plt.xlabel('Time (min)')
plt.show()
def chooseValueOrDefault(str_input, default):
if str_input == "":
return default
else:
return float(str_input)
if __name__ == "__main__":
data, filepath_confirm = getInputLoadFile()
fullOutputPath = getOutputPath()
print("")
print("Please choose settings for the peak detection algorithm. For default values press return")
thresh_str = get_user_input('\tMinimum peak amplitude (default = .02):')
thresh = chooseValueOrDefault(thresh_str,.02)
offset_str = get_user_input('\tOffset (default = 1): ')
offset = chooseValueOrDefault(offset_str,1)
start_WT_str = get_user_input('\tMax rise time (s) (default = 4): ')
start_WT = chooseValueOrDefault(start_WT_str,4)
end_WT_str = get_user_input('\tMax decay time (s) (default = 4): ')
end_WT = chooseValueOrDefault(end_WT_str,4)
settings_dict = {'threshold':thresh,
plt.xlim([0,time_m[-1]])
plt.ylim([data_min-.1,data_max+.1])
plt.title('Motion with Detected "Steps" marked')
plt.ylabel('g')
if x_seconds:
plt.xlabel('Time (s)')
else:
plt.xlabel('Time (min)')
plt.show()
if __name__ == "__main__":
print "This script will extract features related to accelerometer data."
data, filepath_confirm = getInputLoadFile()
output_path = getOutputPath()
time_frames = inputTimeFrames()
features, steps, motion = computeAllAccelerometerFeatures(data, time_frames)
data["steps"] = steps
data["motion"] = motion
saveFeaturesToFile(features, time_frames, output_path)
print ""
plot_ans = raw_input("Do you want to plot the detected steps? (y/n): ")
if 'y' in plot_ans: