def test_block_weights(self):
regimes = np.ones(25)
regimes[range(10, 20)] = 2
regimes[range(21, 25)] = 3
regimes = np.array([
1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 2., 2., 2., 2., 2., 2., 2.,
2., 2., 2., 1., 3., 3., 3., 3.
])
w = pysal.block_weights(regimes)
ww0 = [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
self.assertEquals(w.weights[0], ww0)
wn0 = [1, 2, 3, 4, 5, 6, 7, 8, 9, 20]
self.assertEquals(w.neighbors[0], wn0)
regimes = ['n', 'n', 's', 's', 'e', 'e', 'w', 'w', 'e']
n = len(regimes)
w = pysal.block_weights(regimes)
wn = {
0: [1],
1: [0],
2: [3],
3: [2],
4: [5, 8],
5: [4, 8],
6: [7],
7: [6],
8: [4, 5]
}
self.assertEquals(w.neighbors, wn)
ids = ['id-%i' % i for i in range(len(regimes))]
w = pysal.block_weights(regimes, ids=np.array(ids))
w0 = {'id-1': 1.0}
self.assertEquals(w['id-0'], w0)
w = pysal.block_weights(regimes, ids=ids)
w0 = {'id-1': 1.0}
self.assertEquals(w['id-0'], w0)
def test_block_weights(self):
regimes = np.ones(25)
regimes[range(10, 20)] = 2
regimes[range(21, 25)] = 3
regimes = np.array([1., 1., 1., 1., 1., 1., 1., 1., 1., 1.,
2., 2., 2., 2., 2., 2., 2., 2., 2., 2., 1., 3., 3.,
3., 3.])
w = pysal.block_weights(regimes)
ww0 = [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]
self.assertEquals(w.weights[0], ww0)
wn0 = [1, 2, 3, 4, 5, 6, 7, 8, 9, 20]
self.assertEquals(w.neighbors[0], wn0)
regimes = ['n', 'n', 's', 's', 'e', 'e', 'w', 'w', 'e']
n = len(regimes)
w = pysal.block_weights(regimes)
wn = {0: [1], 1: [0], 2: [3], 3: [2], 4: [5, 8], 5: [4, 8],
6: [7], 7: [6], 8: [4, 5]}
self.assertEquals(w.neighbors, wn)
ids = ['id-%i'%i for i in range(len(regimes))]
w = pysal.block_weights(regimes, ids=np.array(ids))
w0 = {'id-1': 1.0}
self.assertEquals(w['id-0'], w0)
w = pysal.block_weights(regimes, ids=ids)
w0 = {'id-1': 1.0}
self.assertEquals(w['id-0'], w0)
def setUp(self):
f = pysal.open(pysal.examples.get_path("mexico.csv"))
vnames = ["pcgdp%d" % dec for dec in range(1940, 2010, 10)]
y = np.transpose(np.array([f.by_col[v] for v in vnames]))
self.y = y[:, 0]
regimes = np.array(f.by_col('hanson98'))
self.w = ps.block_weights(regimes)
def setUp(self):
f = pysal.open(pysal.examples.get_path("mexico.csv"))
vnames = ["pcgdp%d" % dec for dec in range(1940, 2010, 10)]
y = np.transpose(np.array([f.by_col[v] for v in vnames]))
self.y = y[:, 0]
regimes = np.array(f.by_col('hanson98'))
self.w = ps.block_weights(regimes)
def rank(observations, w=None, regimes=None, method="tau"):
originalObs = observations
observations = observations.transpose()
w = None
a = None
if regimes == None: # generate regimes endogeneously.
from GSA import Regionalization
#TODO fix double transpose issue. fixed
a = Regionalization.generateRegimes(w, originalObs)
w = pysal.block_weights(a)
else: # regimes provided by user.
w = pysal.block_weights(regimes)
if method == "tau":
result = [pysal.SpatialTau(observations[:, i], observations[:, i + 1], w, 100) for i in
range(0, len(w.id_order))]
ret = [(r.tau_spatial, r.taus.mean(), r.tau_spatial_psim) for r in result]
else: # method == "theta"
if regimes == None:
result = pysal.Theta(regime=a, y=observations)
else:
result = pysal.Theta(regime=regimes, y=observations)
ret = [(r.theta, r.pvalue_left, r.pvalue_right) for r in result]
return ret
def bounded_world(r, c, nr, nc):
'''
Create W object for a bounded neighborhood topology based on grids (pixels
and neighborhoods)
NOTE: if r/nr or c/nc are not natural, the number of neighborhoods can be
slightly inacurate
...
Arguments
---------
r : int
Number of pixels on the Y axis (rows)
c : int
Number of pixels on the X axis (columns)
nr : int
Number of neighborhoods on the Y axis (rows)
nc : int
Number of neighborhoods on the X axis (columns)
Returns
-------
W : pysal.W
Weights object
ns : ndarray
Cardinalities for every observation to a neighborhood
xys : ndarray
Nx2 array with coordinates of pixels
'''
x, y = np.indices((r, c))
ir = int(np.round((r * 1.) / nr))
ic = int(np.round((c * 1.) / nc))
nrb = [i + ir for i in range(-ir, r, ir)]
ncb = [i + ic for i in range(-ic, c, ic)]
world = np.zeros((r, c), dtype=int)
n = 0
for i in range(len(nrb) - 1):
for j in range(len(ncb) - 1):
world[nrb[i]:nrb[i + 1], ncb[j]:ncb[j + 1]] = n
n += 1
world = world.flatten()
w = ps.block_weights(world)
return w, world, np.hstack((x.flatten()[:, None], y.flatten()[:, None]))
def bounded_world(r, c, nr, nc):
'''
Create W object for a bounded neighborhood topology based on grids (pixels
and neighborhoods)
NOTE: if r/nr or c/nc are not natural, the number of neighborhoods can be
slightly inacurate
...
Arguments
---------
r : int
Number of pixels on the Y axis (rows)
c : int
Number of pixels on the X axis (columns)
nr : int
Number of neighborhoods on the Y axis (rows)
nc : int
Number of neighborhoods on the X axis (columns)
Returns
-------
W : pysal.W
Weights object
ns : ndarray
Cardinalities for every observation to a neighborhood
xys : ndarray
Nx2 array with coordinates of pixels
'''
x, y = np.indices((r, c))
ir = int(np.round((r*1.)/nr))
ic = int(np.round((c*1.)/nc))
nrb = [i+ir for i in range(-ir, r, ir)]
ncb = [i+ic for i in range(-ic, c, ic)]
world = np.zeros((r, c), dtype=int)
n = 0
for i in range(len(nrb)-1):
for j in range(len(ncb)-1):
world[nrb[i]: nrb[i+1], ncb[j]:ncb[j+1]] = n
n += 1
world = world.flatten()
w = ps.block_weights(world)
return w, world, np.hstack((x.flatten()[:, None], y.flatten()[:, None]))
