def GenerateMacroActionsFromeFile(batch_size):
filename = '../src/taxi18.dom'
m = GenerateRoadModelFromFile(filename)
start_location = np.array([m.GetRandomStartLocation(batch_size)])
current_location = start_location[-1, :]
print current_location
print m.GenerateAllRoadMacroActions(current_location, batch_size)[0].shape
def CheckNeighboursTest():
file_name = '../src/taxi18.dom'
m = GenerateRoadModelFromFile(file_name)
locs = m.locations
for loc in locs:
loc = tuple(loc)
# length = len(m.GenerateRoadMacroActions(tuple(loc), 3))
neighbours = m.GetNeighbours(loc)
# print loc, neighbours
for nei in neighbours:
n_neighbours = m.GetNeighbours(nei)
if loc in n_neighbours:
print loc, nei
def IterateOverMacroActions(batch_size):
file_name = '../datasets/slot18/tlog18.dom'
m = GenerateRoadModelFromFile(file_name)
# m.AddTwoSidedRoads()
locs = m.locations
sum = 0.0
max = 0
count = 0
for loc in locs:
length = len(m.GenerateAllRoadMacroActions(tuple(loc), batch_size))
sum += length
if length > max:
max = length
# count the number of locations where we have macroactions
if length > 0:
# print loc, length
count += 1
print count, max, sum / count
"""
def HistTests():
file_name = '../src/taxi44.dom'
m = GenerateRoadModelFromFile(file_name)
hypers_file = open('hypers44.txt', 'w')
locs = m.locations
vals = m.values
for i in range(locs.shape[0]):
if vals[i] != -1.0:
loc = locs[i, :]
hypers_file.write(
str(loc[0]) + ' ' + str(loc[1]) + ' ' + str(vals[i]) + '\n')
hypers_file.close()
list_vals = [v for v in vals.tolist() if v != -1.0]
updated_vals = np.asarray(list_vals)
print scipy.stats.skew(updated_vals)
plt.hist(updated_vals, bins=50)
plt.show()
def LogTransformTest():
file_name = '../src/taxi18.dom'
m = GenerateRoadModelFromFile(file_name)
vals = m.values
for val in vals:
print val
def NewDatasetTest():
file_name = '../datasets/slot18/tlog18.dom'
m = GenerateRoadModelFromFile(file_name)
print m.mean
def GetMacroActionsOfLocation(loc, batch_size):
filename = '../src/taxi18.dom'
m = GenerateRoadModelFromFile(filename)
return m.GenerateAllRoadMacroActions(loc, batch_size)
def LoadActionsFromFileTest(folder_name, batch_size):
filename = '../datasets/slot18/taxi18.dom'
m = GenerateRoadModelFromFile(filename)
m.LoadSelectedMacroactions(folder_name, batch_size)
error = 0
for test_p in test:
point = locs[test_p: test_p + 1, :]
assert len(point.shape) == 2
weights = gp.GPWeights(locations=locs_train, current_location=point, cholesky=cholesky)
mu_predict = gp.GPMean(measurements=vals_train, weights=weights)
truth = vals[test_p]
print point, truth, mu_predict
error += (mu_predict- truth)**2
print sqrt(error / (test_number_of_points))
if __name__ == "__main__":
file_name = '../datasets/slot18/tlog18.dom'
m = GenerateRoadModelFromFile(file_name)
allowed_indexes = m.informative_locations_indexes
locs = m.locations[allowed_indexes, :]
vals = m.values[allowed_indexes]
#TestPrediction(locs, vals)
pairs = GenerateGridPairs(range(8,12), range(40,44)).tolist()
raw_indexes = map(lambda x: TupleToLine(x, m.dim_1, m.dim_2), pairs)
indexes = [x for x in raw_indexes if vals[x]!= m.NO_DATA_CONST]
selected_locs = locs[indexes, :]
print selected_locs
selected_vals = vals[indexes]
TestPrediction(selected_locs, selected_vals)