def data_visualization(K, dynamic_datasets_path):
print("Going to visualize bus-trajectories..")
dataSets = readDatasets.read_dataset(True, False, False,
dynamic_datasets_path)
trainSet = dataSets[0]
#print trainSet.shape[0] # DEBUG!
#print trainSet['Trajectory'] # DEBUG!
journeyPatternIDs, trainTrajs, trainListSize = TrainData.getListsOfTrainData(
trainSet)
storeMapsDir = os.path.join('..', '..', 'Resources', 'maps', 'task1')
if not os.path.isdir(storeMapsDir):
os.makedirs(storeMapsDir)
selectedPatternIDs = []
numOfSelectedPatterns = 0
start_time = time.time()
maxSecondsToWait = 120
while True:
if numOfSelectedPatterns == K:
print('Finished plotting ' + K.__str__() +
' distinct random patterns.')
break
elif (time.time() - start_time) > maxSecondsToWait:
print('The program could not find ' + K.__str__() +
' distinct random patterns in the specified time: ' +
maxSecondsToWait.__str__())
break
randomTrain = random.randint(0, trainListSize - 1)
curPatternID = journeyPatternIDs[randomTrain]
if curPatternID not in selectedPatternIDs:
selectedPatternIDs.append(curPatternID)
# plot the new pattern
print('Going to plot a new random train..')
longitudes, latitudes = GetCoordinates.getCoordinates(
trainTrajs[randomTrain])
fileName = "train" + randomTrain.__str__(
) + "_Pattern_" + curPatternID + ".html"
GmPlot.gmPlot(latitudes, longitudes,
os.path.join(storeMapsDir, fileName))
numOfSelectedPatterns += 1
def data_visualization(K):
dataSets = readDatasets.read_dataset(True, False, False)
trainSet = dataSets[0]
#print trainSet.shape[0] # DEBUG!
#print trainSet['Trajectory'] # DEBUG!
journeyPatternIDs, trainTrajs, trainListSize = TrainData.getListsOfTrainData(
trainSet)
storeMapsDir = "../../Resources/maps/task1"
if not os.path.isdir(storeMapsDir):
os.makedirs(storeMapsDir)
selectedPatternIDs = []
numOfSelectedPatterns = 0
start_time = time.time()
maxSecondsToWait = 120
while True:
if numOfSelectedPatterns == K:
print 'Finished plotting ' + K.__str__(
) + ' distinct random patterns.'
break
elif (time.time() - start_time) > maxSecondsToWait:
print 'The program could not find ' + K.__str__(
) + ' distinct random patterns in the specified time: ' + maxSecondsToWait.__str__(
)
break
randomTrain = random.randint(0, trainListSize - 1)
curPatternID = journeyPatternIDs[randomTrain]
if curPatternID not in selectedPatternIDs:
selectedPatternIDs.append(curPatternID)
# plot the new pattern
print 'Going to plot a new random train..'
longtitutes, latitudes = GetCoordinates.getCoordinates(
trainTrajs[randomTrain])
GmPlot.gmPlot(
latitudes, longtitutes, storeMapsDir + "/train" +
randomTrain.__str__() + "_Pattern_" + curPatternID + ".html")
numOfSelectedPatterns += 1
def runLCSS(K, dynamic_datasets_path):
print('\nLCSS start..')
dataSets = readDatasets.read_dataset(True, False, True,
dynamic_datasets_path)
trainSet = dataSets[0]
testSetA2 = dataSets[1]
journeyPatternIDs, trainTrajs, trainListSize = TrainData.getListsOfTrainData(
trainSet)
storeMapsDir = os.path.join('..', '..', 'Resources', 'maps', 'task2A2')
if not os.path.isdir(storeMapsDir):
os.makedirs(storeMapsDir)
kMaxs = KMaxs(K)
start_time = time.time()
lastTime = start_time # For in-the-middle elapsed-time.
testNum = 0
for trajectoryTest in testSetA2['Trajectory']:
# print trajectoryTest
testNum += 1
# if testNum <= 2:
# continue
# if testNum == 2:
# break
nearestNeighbors = []
sorted_subSeqsSizes = []
print('\nChecking for ' + K.__str__() +
' longest common sub-sequences of test ' + testNum.__str__())
# print trajectoryTest # DEBUG!
iterations = 0
sorted_subSequences = []
for i in range(
0,
trainListSize): # IDs and Trajectories are of the same size.
# print i
trajectoryTrain = trainTrajs[i]
LongestCS = lcs(trajectoryTrain, trajectoryTest)
if not LongestCS:
continue
# else:
# print i.__str__() + ') LCS length: ' + len(LongestCS).__str__()
# if iterations == 10:
# break
# else:
# iterations += 1
kMaxs.checkMinLengthAndInsert([i, LongestCS, len(LongestCS)])
curTime = time.time()
curElapsedTime = curTime - lastTime
lastTime = curTime
print('\nTest: ' + testNum.__str__() + ') finished in ' +
time.strftime("%H:%M:%S", time.gmtime(curElapsedTime)))
# Plot test
full_longitudes, full_latitudes = GetCoordinates.getCoordinates(
trajectoryTest)
fileName = "lcss" + testNum.__str__(
) + "-test-Time(sec)_" + curElapsedTime.__str__() + ".html"
GmPlot.gmPlot(full_latitudes,
full_longitudes,
os.path.join(storeMapsDir, fileName),
zoom=13)
# So now we pic the top 5 and we plot them....
sorted_subSequences = sorted(kMaxs.getArrayList(),
reverse=True,
key=lambda tup: tup[2])
kMaxs.resetArrayList(
) # Reset arrayList before going to the next testSet.
for i in range(0, len(sorted_subSequences)):
if i == 5: break
print("Train " + sorted_subSequences[i][0].__str__() + ") PatternID: " \
+ journeyPatternIDs[sorted_subSequences[i][0]].__str__() \
+ ", MatchingPoints: " + sorted_subSequences[i][2].__str__() + ".html")
curSubSeqTrajectory = trainTrajs[sorted_subSequences[i][0]]
full_longitudes, full_latitudes = GetCoordinates.getCoordinates(
curSubSeqTrajectory)
curSubSeqTrajectory = sorted_subSequences[i][1]
sub_longitudes, sub_latitudes = GetCoordinates.getCoordinates(
curSubSeqTrajectory)
fileName = "lcss" + testNum.__str__() + "-train" \
+ sorted_subSequences[i][0].__str__() + "_PatternID_" \
+ journeyPatternIDs[sorted_subSequences[i][0]].__str__() \
+ "-MatchingPoints_" + sorted_subSequences[i][2].__str__() + ".html"
GmPlot.gmPlotOfColours(full_latitudes, full_longitudes,
sub_latitudes, sub_longitudes,
os.path.join(storeMapsDir, fileName))
print("\nElapsed time of KNNwithLCSS for 'test_set_a2': ",
time.strftime("%H:%M:%S", time.gmtime(time.time() - start_time)),
'mins')
def findKnearestNeighbors(K, maxWarpingWindowPercentage, plotPatterns,
makeListOfAllNeighbors, trainSet, testSet):
print('KNN-with-DTW starts..')
journeyPatternIDs, trainTrajs, trainListSize = TrainData.getListsOfTrainData(
trainSet)
if plotPatterns:
storeMapsDir = "../../Resources/maps/task2A1"
if not os.path.isdir(storeMapsDir):
os.makedirs(storeMapsDir)
dtw = Dtw(max_warping_window_percentage=maxWarpingWindowPercentage)
kMins = KMins(K)
sorted_nearestNeighbors_forAllTests = []
inf_costs_count = 0
testNum = 0
start_time = time.time()
lastTime = start_time # For in-the-middle elapsed-time.
for trajectoryTest in testSet['Trajectory']:
inf_costs_count = 0
testNum += 1
# if testNum != 2:
# continue
# if testNum <= 4:
# continue
# if testNum == 3:
# print 'InfCount: ' + inf_costs_count.__str__()
# break
print('\nChecking for ' + K.__str__() + ' nearest-neighbors of test ' +
testNum.__str__() + '..')
# print trajectoryTest # DEBUG!
min_cost = 1000
minJourneyPatternId = ''
min_i = 0
for i in range(
0,
trainListSize): # IDs and Trajectories are of the same size.
# if i == 1500:
# break
curPatternID = journeyPatternIDs[i]
trajectoryTrain = trainTrajs[i]
cost = dtw._dtw_distance(trajectoryTest, trajectoryTrain)
if math.isinf(cost):
# diff = abs(len(trajectoryTrain) - len(trajectoryTest))
# print 'returned inf while having a diff of: ' + diff.__str__()
#print 'We have an inf cost in DTW: test:' + testNum.__str__() + ' - trainPatternID: ' + curPatternID
inf_costs_count += 1
continue
kMins.checkMaxCostAndInsert([i, curPatternID, cost])
# if not makeListOfAllNeighbors and cost <= 25.5:
# print testNum.__str__() + '-' + i.__str__() + ') Cost (lessOrEqual to 25.5): ' + cost.__str__()\
# + ' from journeyPatternID: ' + minJourneyPatternId.__str__()
if cost < min_cost:
min_cost = cost
min_i = i
minJourneyPatternId = curPatternID
print(testNum.__str__() + '-' + min_i.__str__() + ') Found new minCost: ' + min_cost.__str__()\
+ ' from journeyPatternID: ' + minJourneyPatternId.__str__())
curTime = time.time()
curElapsedTime = curTime - lastTime
lastTime = curTime
sorted_nearestNeighbors = sorted(kMins.getArrayList(),
key=lambda tup: tup[2])
kMins.resetArrayList(
) # Reset arrayList before going to the next testSet.
print('\nTest: ' + testNum.__str__() + ') finished in ' + time.strftime("%H:%M:%S", time.gmtime(curElapsedTime))\
+ '\nMax warping window percentage: ' + maxWarpingWindowPercentage.__str__() + '\n\'Inf\' costs found in this test: ' + inf_costs_count.__str__() \
+ '\nMin_journeyPatternId: ' + minJourneyPatternId.__str__() + ' Min_i: ' + min_i.__str__() + ' Min_cost: ' + min_cost.__str__() \
+ '\nSorted mins: ')
if plotPatterns:
# Plot test
longtitutes, latitudes = GetCoordinates.getCoordinates(
trajectoryTest)
GmPlot.gmPlot(latitudes,
longtitutes,
storeMapsDir + "/dtw" + testNum.__str__() +
"-test-Time(sec)_" + curElapsedTime.__str__() +
".html",
zoom=13)
sorted_nearestNeighbors_forTest = []
# Plot trains
for i in range(0, len(sorted_nearestNeighbors)):
print("i: ", sorted_nearestNeighbors[i][0], ", PatternID: ", sorted_nearestNeighbors[i][1]\
, ", DTW-cost: ", sorted_nearestNeighbors[i][2])
# Make a list with all the neighbours
if makeListOfAllNeighbors:
sorted_nearestNeighbors_forTest.append(
sorted_nearestNeighbors[i][1])
if plotPatterns:
curTrainTrajectory = trainTrajs[sorted_nearestNeighbors[i][0]]
longtitutes, latitudes = GetCoordinates.getCoordinates(
curTrainTrajectory)
GmPlot.gmPlot(
latitudes,
longtitutes,
storeMapsDir + "/dtw" + testNum.__str__() + "-train" +
sorted_nearestNeighbors[i][0].__str__() + "-PatternID_" +
sorted_nearestNeighbors[i][1].__str__() + "-DTW_" +
sorted_nearestNeighbors[i][2].__str__() + ".html",
zoom=13)
if makeListOfAllNeighbors:
# Make a list with all the neighbours for all the tests
sorted_nearestNeighbors_forAllTests.append(
sorted_nearestNeighbors_forTest)
print("\nElapsed time of KNNwithDTW for 'test_set_a1': ",
time.strftime("%H:%M:%S", time.gmtime(time.time() - start_time)))
return sorted_nearestNeighbors_forAllTests