def getAnglePlatform(x_pos, y_pos):
with open('fuzzycontrol_normal_3.pkl', 'rb') as f:
anf = pickle.load(f)
input_val = np.array([[x_pos, y_pos]])
anfis_matrix = anfis.predict(anf, input_val)
return np.array([anfis_matrix[0,0], anfis_matrix[0, 1], anfis_matrix[0, 2]])
def fuzzy_error_test(anf, test_table_csv):
# grab the input positions from the test table
test_xyz = numpy.loadtxt(test_table_csv, delimiter=',', usecols=[0, 1, 2])
# grab the expected angles
test_angles = numpy.loadtxt(test_table_csv,
delimiter=',',
usecols=[3, 4, 5])
# with open('fuzzy_test_gauss.pkl', 'rb') as f:
# anf = pickle.load(f)
# initialize the total error count
total_error = 0
average_error = []
# loop through each row in the demo table
for row_num in range(len(test_xyz)):
# grab the input value for ANFIS
input_val = test_xyz[row_num]
# put into a numpy array
input_val = numpy.array([input_val])
# evaluate ANFIS and get the predicted output
predicted_output = anfis.predict(anf, input_val)
# calculate the percent error between the predicted and expected
percent_error = numpy.mean(
abs(test_angles[row_num] - predicted_output) /
test_angles[row_num])
# add onto the total error
total_error = total_error + percent_error
average_error.append(percent_error)
# print for each row evaluated
print row_num, len(test_xyz)
return total_error, numpy.mean(average_error)
pickle_filename = []
pickle_filename.append(log_filename[:-4])
pickle_filename.append('.pkl')
pickle_filename = ''.join(pickle_filename)
with open(pickle_filename, 'wb') as f:
pickle.dump(anf, f, pickle.HIGHEST_PROTOCOL)
logging.info("Wrote the PICKLE FILE: %s", pickle_filename)
print time.time() - t
total_error, average_error = fuzzy_error_test(anf, "test_table_176_184.csv")
logging.info("The total error is: %f", total_error)
logging.info("The average error is: %f", average_error)
print "The total error is: ", total_error
print "The average error is: ", average_error
# var = numpy.array([[3, 4], [3, 4]])
var = np.array([[0,0,-207]])
print "input", var
# print the predicted value based on the trained set
print "The answer is", anfis.predict(anf, var)
except:
logging.exception('An exception has occured!!')
raise
# set the input value input_val = numpy.array([[0, 0, -207.70890486488082]]) # set what the expected output should be output_val = numpy.array([175, 175, 175]) print "input", input_val # print "the length is", len(var[:,0]) # # var = numpy.asanyarray(var) # print "the shape", var.shape # print "the shape index", var.shape[0] # print "the shape index", var.shape[1] # start the timer start = timeit.timeit() # print the predicted value based on the trained set anfis_ans = anfis.predict(anf, input_val) print "The answer is", anfis_ans # stop the timer end = timeit.timeit() print "Time: ", end - start # calculate the error mse = ((output_val - anfis_ans)**2).mean(axis=None) print "Error: ", mse * 100