def test_build_ps_owa_factory():
import os
csv_file = os.path.join(os.path.dirname(__file__), "iris.csv")
data = np.genfromtxt(csv_file, dtype=float, delimiter=',', names=True)
X = np.array([data["sepallength"], data["sepalwidth"], data["petallength"], data["petalwidth"]]).T
y = data["class"]
from sklearn.preprocessing import MinMaxScaler
X = MinMaxScaler().fit_transform(X)
l = nfpc.FuzzyPatternClassifier(
membership_factory=t_factory,
aggregation_factory=nfpc.GAOWAFactory(optimizer=nfpc.ps_owa_optimizer())
)
from sklearn.model_selection import cross_val_score
scores = cross_val_score(l, X, y, cv=10)
mean = np.mean(scores)
print("mean", mean)
assert 0.92 < mean
def test_simple_fpc():
iris = datasets.load_iris()
X = iris.data
y = iris.target
c = nfpc.FuzzyPatternClassifier()
c.fit(X, y)
mu = nfpc.predict_protos(X, c.protos_, c.aggregation_)
pro = c.predict_proba(X)
pro_sum = pro.sum(axis=1)
assert np.all(pro_sum > 0.)
def test_build_meowa_factory():
iris = datasets.load_iris()
X = iris.data
y = iris.target
from sklearn.preprocessing import MinMaxScaler
X = MinMaxScaler().fit_transform(X)
l = nfpc.FuzzyPatternClassifier(membership_factory=t_factory,
aggregation_factory=nfpc.MEOWAFactory())
from sklearn.model_selection import cross_val_score
scores = cross_val_score(l, X, y, cv=10)
mean = np.mean(scores)
assert 0.80 < mean
