def compute_statistics()

in Experiments/MIME_DataLoader.py [0:0]

• 51 lines of code
• 1 McCabe index (conditional complexity)

	def compute_statistics(self):

		self.state_size = 16
		self.total_length = self.__len__()
		mean = np.zeros((self.state_size))
		variance = np.zeros((self.state_size))
		mins = np.zeros((self.total_length, self.state_size))
		maxs = np.zeros((self.total_length, self.state_size))
		lens = np.zeros((self.total_length))

		# And velocity statistics. 
		vel_mean = np.zeros((self.state_size))
		vel_variance = np.zeros((self.state_size))
		vel_mins = np.zeros((self.total_length, self.state_size))
		vel_maxs = np.zeros((self.total_length, self.state_size))

		
		for i in range(self.total_length):

			print("Phase 1: DP: ",i)
			data_element = self.__getitem__(i)

			if data_element['is_valid']:
				demo = data_element['demo']
				vel = np.diff(demo,axis=0)
				mins[i] = demo.min(axis=0)
				maxs[i] = demo.max(axis=0)
				mean += demo.sum(axis=0)
				lens[i] = demo.shape[0]

				vel_mins[i] = abs(vel).min(axis=0)
				vel_maxs[i] = abs(vel).max(axis=0)
				vel_mean += vel.sum(axis=0)			

		mean /= lens.sum()
		vel_mean /= lens.sum()

		for i in range(self.total_length):

			print("Phase 2: DP: ",i)
			data_element = self.__getitem__(i)
			
			# Just need to normalize the demonstration. Not the rest. 
			if data_element['is_valid']:
				demo = data_element['demo']
				vel = np.diff(demo,axis=0)
				variance += ((demo-mean)**2).sum(axis=0)
				vel_variance += ((vel-vel_mean)**2).sum(axis=0)

		variance /= lens.sum()
		variance = np.sqrt(variance)

		vel_variance /= lens.sum()
		vel_variance = np.sqrt(vel_variance)

		max_value = maxs.max(axis=0)
		min_value = mins.min(axis=0)

		vel_max_value = vel_maxs.max(axis=0)
		vel_min_value = vel_mins.min(axis=0)

		np.save("MIME_Orig_Mean.npy", mean)
		np.save("MIME_Orig_Var.npy", variance)
		np.save("MIME_Orig_Min.npy", min_value)
		np.save("MIME_Orig_Max.npy", max_value)
		np.save("MIME_Orig_Vel_Mean.npy", vel_mean)
		np.save("MIME_Orig_Vel_Var.npy", vel_variance)
		np.save("MIME_Orig_Vel_Min.npy", vel_min_value)
		np.save("MIME_Orig_Vel_Max.npy", vel_max_value)