def test_density_features(self):
df = DensityFeatures()
f = df.featurize(self.diamond)
self.assertAlmostEqual(f[0], 3.49, 2)
self.assertAlmostEqual(f[1], 5.71, 2)
self.assertAlmostEqual(f[2], 0.25, 2)
f = df.featurize(self.nacl)
self.assertAlmostEqual(f[0], 2.105, 2)
self.assertAlmostEqual(f[1], 23.046, 2)
self.assertAlmostEqual(f[2], 0.620, 2)
def test_density_features(self):
df = DensityFeatures()
f = df.featurize(self.diamond)
self.assertAlmostEqual(f[0], 3.49, 2)
self.assertAlmostEqual(f[1], 5.71, 2)
self.assertAlmostEqual(f[2], 0.25, 2)
f = df.featurize(self.nacl)
self.assertAlmostEqual(f[0], 2.105, 2)
self.assertAlmostEqual(f[1], 23.046, 2)
self.assertAlmostEqual(f[2], 0.620, 2)
def test_density_features(self):
df = DensityFeatures()
f = df.featurize(self.diamond)
self.assertAlmostEqual(f[0], 3.49, 2)
self.assertAlmostEqual(f[1], 5.71, 2)
self.assertAlmostEqual(f[2], 0.25, 2)
f = df.featurize(self.nacl)
self.assertAlmostEqual(f[0], 2.105, 2)
self.assertAlmostEqual(f[1], 23.046, 2)
self.assertAlmostEqual(f[2], 0.620, 2)
nacl_disordered = copy.deepcopy(self.nacl)
nacl_disordered.replace_species({"Cl1-": "Cl0.99H0.01"})
self.assertFalse(df.precheck(nacl_disordered))
structures = [self.diamond, self.nacl, nacl_disordered]
df2 = pd.DataFrame({"structure": structures})
self.assertAlmostEqual(df.precheck_dataframe(df2, "structure"), 2 / 3)
def test_density_features(self):
df = DensityFeatures()
f = df.featurize(self.diamond)
self.assertAlmostEqual(f[0], 3.49, 2)
self.assertAlmostEqual(f[1], 5.71, 2)
self.assertAlmostEqual(f[2], 0.25, 2)
f = df.featurize(self.nacl)
self.assertAlmostEqual(f[0], 2.105, 2)
self.assertAlmostEqual(f[1], 23.046, 2)
self.assertAlmostEqual(f[2], 0.620, 2)
nacl_disordered = copy.deepcopy(self.nacl)
nacl_disordered.replace_species({"Cl1-": "Cl0.99H0.01"})
self.assertFalse(df.precheck(nacl_disordered))
structures = [self.diamond, self.nacl, nacl_disordered]
df2 = pd.DataFrame({"structure": structures})
self.assertAlmostEqual(df.precheck_dataframe(df2, "structure"), 2 / 3)
def _extract_features(self, df_input):
"""
Extract features using Matminer from the 'structure' column in
df_input
Args:
df_input (DataFrame): Pandas DataFrame whcih conatains features
from Materials Project Database of the input samples
Returns:
df_extracted (DataFrame): Pandas DataFrame which contains
features of input samples extracted using Matminer
"""
# Dropping the 'theoretical' column
df_input.drop(columns=["theoretical"], inplace=True)
# Extracting the features
dfeat = DensityFeatures()
symmfeat = GlobalSymmetryFeatures()
mfeat = Meredig()
cefeat = CohesiveEnergy()
df_input["density"] = df_input.structure.apply(
lambda x: dfeat.featurize(x)[0])
df_input["vpa"] = df_input.structure.apply(
lambda x: dfeat.featurize(x)[1])
df_input["packing fraction"] = df_input.structure.apply(
lambda x: dfeat.featurize(x)[2])
df_input["spacegroup_num"] = df_input.structure.apply(
lambda x: symmfeat.featurize(x)[0])
df_input["cohesive_energy"] = df_input.apply(
lambda x: cefeat.featurize(
x.structure.composition,
formation_energy_per_atom=x.formation_energy_per_atom,
)[0],
axis=1,
)
df_input["mean AtomicWeight"] = df_input.structure.apply(
lambda x: mfeat.featurize(x.composition)[-17])
df_input["range AtomicRadius"] = df_input.structure.apply(
lambda x: mfeat.featurize(x.composition)[-12])
df_input["mean AtomicRadius"] = df_input.structure.apply(
lambda x: mfeat.featurize(x.composition)[-11])
df_input["range Electronegativity"] = df_input.structure.apply(
lambda x: mfeat.featurize(x.composition)[-10])
df_input["mean Electronegativity"] = df_input.structure.apply(
lambda x: mfeat.featurize(x.composition)[-9])
# Drop 'structure' column
df_input.drop(columns=["structure"], inplace=True)
# ignore compounds that failed to featurize
df_extracted = df_input.fillna(
df_input.mean()).query("cohesive_energy > 0.0")
# Re-arranging the 'PU Label' column
pu_label = df_extracted["PU_label"]
df_extracted = df_extracted.drop(["PU_label"], axis=1)
df_extracted["PU_label"] = pu_label
# Drop the icsd_ids column
df_extracted.drop(columns=["icsd_ids"], inplace=True)
return df_extracted