ydir = 0
if util.includeVelocity == True:
features.append(yvel)
if util.includeAcceleration == True:
features.append(yacc)
if util.includeHeading == True:
features.append(heading)
if util.includeYDirection == True:
features.append(ydir)
tdX = np.array(util.removeColFromRow(features, 0))
tdY = np.array(util.removeColFromRow(features, 1))
predY = neighY.predict(tdY)
actual = test1Data[i + offset][1]
prediction = predY[0]
actuals.append(actual)
predictions.append(prediction)
# errors.append(util.error([actual], [prediction]))
if onlyTrainingAndCV == False:
# util.plotLines(actuals, predictions, 'Actual position', 'Predicted position')
# util.plotGraph(actuals, predictions, 'Actual position', 'Predicted position')
util.plotLine(actuals, 'Actual position')
util.plotLine(predictions, 'Predicted position')
# util.plotLine(errors, 'Error graph')
print len(actuals), len(predictions)
# print np.sum(errors)
else:
util.plotLine(cvScoresY, 'CV label y')
#parse the input txt,eg:text01.txt
givenData = parseTxt(filename)
#use the last 60 frames of gieven input, eg test01.txt as actual
actual = givenData[len(givenData) - 60:]
# feed the program input data with last 60 frames chopped off
allData = consolidateAllData()
output = particleFilterPrediction(givenData[:len(givenData) - 60],
allData)
#####need to write output to txt#########
#use utility to test our error and make plots
errors = []
for i in range(0, len(output)):
#errors.append(util.error([actual[i]],[output[i]]))
ok = ((output[i][0] - actual[i][0])**2) + (
(output[i][1] - actual[i][1])**2)
errors.append(ok)
util.plotGraph(actual, output, 'Actual position', 'Predicted position')
util.plotLine(errors, 'Error graph')
print len(actual), len(output)
print "L2 error is >>>>>>>>"
print math.sqrt(np.sum(errors))
else:
sys.exit(1)
test1 = open('inputs/test01.txt')
test1Data = np.loadtxt(test1, delimiter=',')
test1Data = util.normalizeData(test1Data)
errors = []
NUM_INPUTS = len(test1Data) - offset
for i in range(1750, NUM_INPUTS):
current = test1Data[i]
features = util.createFeatureRow(test1Data, i, offset, current)
td = np.array(features)
predX = neigh.predict(td)
predY = neighY.predict(td)
actual = test1Data[i + offset]
prediction = [predX[0], predY[0]]
actuals.append(actual)
predictions.append(prediction)
errors.append(util.error([actual], [prediction]))
if onlyTrainingAndCV == False:
# util.plotLines(actuals, predictions, 'Actual position', 'Predicted position')
# util.plotGraph(actuals, predictions, 'Actual position', 'Predicted position')
util.plotData(actuals, 'Actual position')
util.plotData(predictions, 'Predicted position')
util.plotLine(errors, 'Error graph')
print len(actuals), len(predictions)
print np.sum(errors)
else:
util.plotLine(cvScoresX, 'CV label x')
util.plotLine(cvScoresY, 'CV label y')
# util.plotLines(cvScoresX, cvScoresY, 'CV label X', 'CV label y')