def test_r_10_diag(self):
import os
import numpy as np
from pyeee import morris_sampling, elementary_effects
nt = 10
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
# Check 1
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal, Diagnostic=1)
self.assertEqual(list(np.around(mat[0,0:5],3)), [0.6, 2.2, 2., 4.6, 8.])
self.assertEqual(list(np.around(vec[0:5],3)), [12., 11., 5., 1., 9.])
self.assertTrue(os.path.exists('morris_diag_new_strategy.png'))
self.assertTrue(os.path.exists('morris_diag_old_strategy.png'))
if os.path.exists('morris_diag_new_strategy.png'): os.remove('morris_diag_new_strategy.png')
if os.path.exists('morris_diag_old_strategy.png'): os.remove('morris_diag_old_strategy.png')
# Check 2
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps, Diagnostic=1)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.485, 0.445, 0.438, 0.58, 0.645])
self.assertEqual(list(np.around(sa[0:5,1],3)), [0.47, 0.502, 0.816, 0.722, 0.418])
self.assertFalse(os.path.exists('morris_diag_new_strategy.png'))
self.assertFalse(os.path.exists('morris_diag_old_strategy.png'))
def test_r_1(self):
import numpy as np
from pyeee import morris_sampling, elementary_effects
nt = 1
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal, Diagnostic=self.Diagnostic)
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.579, 0.009, 0.239, 0.864, 0.876])
self.assertEqual(list(np.around(sa[0:5].squeeze(),3)), [-0.579, -0.009, -0.239, -0.864, 0.876])
def test_r_10_nan(self):
import numpy as np
from pyeee import morris_sampling, elementary_effects
nt = 10
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
out[1:nt*self.nparam:self.nparam//2] = np.nan
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal, Diagnostic=self.Diagnostic)
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.368, 0.309, 0.549, 0.534, 0.65])
self.assertEqual(list(np.around(sa[~np.isnan(sa[:,1]),1],3)),
[0.47, 0.816, 0.722, 0.418, -0.653, -0.941, 0.863, -1.265, -0.424, -0.786, 0.183])
def test_groups_nan(self):
import numpy as np
from pyeee import morris_sampling, elementary_effects
ngroup = 5
Groups = np.random.randint(0, 4, (self.nparam,ngroup))
nt = 10
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
out[1:nt*self.nparam:self.nparam//2] = np.nan
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal,
GroupMat=Groups, Diagnostic=self.Diagnostic)
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps, Group=Groups)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.49, 0.425, 0.427, 0.441, 0.441])
self.assertEqual(list(np.around(sa[0:5,1],3)), [0.279, 0.557, 0.557, 0.557, 0.557])
def test_r_10(self):
import numpy as np
from pyeee import morris_sampling, elementary_effects
nt = 10
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
# Check 1
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal, Diagnostic=self.Diagnostic)
self.assertEqual(list(np.around(mat[0,0:5],3)), [0.6, 2.2, 2., 4.6, 8.])
self.assertEqual(list(np.around(vec[0:5],3)), [12., 11., 5., 1., 9.])
# Check 2
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.485, 0.445, 0.438, 0.58, 0.645])
self.assertEqual(list(np.around(sa[0:5,1],3)), [0.47, 0.502, 0.816, 0.722, 0.418])
def test_groups(self):
import numpy as np
from pyeee import morris_sampling, elementary_effects
ngroup = 5
Groups = np.random.randint(0, 4, (self.nparam,ngroup))
nt = 10
nsteps = 6
ntotal = 100
out = np.random.random(nt*(self.nparam+1))
# Check 1
mat, vec = morris_sampling(self.nparam, self.LB, self.UB, nt, nsteps=nsteps, ntotal=ntotal,
GroupMat=Groups, Diagnostic=self.Diagnostic)
self.assertEqual(list(np.around(mat[0,0:5],3)), [0.2, 1.8, 3.8, 7., 8.])
self.assertEqual(list(np.around(vec[0:5],3)), [3., 0., 1., 4., 2.])
# Check 2
sa, res = elementary_effects(self.nparam, mat, vec, out, nsteps=nsteps, Group=Groups)
self.assertEqual(list(np.around(res[0:5,0],3)), [0.531, 0.43, 0.432, 0.443, 0.443])
self.assertEqual(list(np.around(sa[0:5,1],3)), [0.279, 0.557, 0.557, 0.557, 0.557])