def test_predict_transform(self):
predict_transformer = FeatureUnion([
("identity", FunctionTransformer(None)),
("log10", FunctionTransformer(numpy.log10))
])
pipeline = PMMLPipeline([("estimator", DummyRegressor())], predict_transformer = predict_transformer)
X = DataFrame([[1, 0], [2, 0], [3, 0]], columns = ["X1", "X2"])
y = Series([0.5, 1.0, 1.5], name = "y")
pipeline.fit(X, y)
y_pred = [1.0, 1.0, 1.0]
y_predt = [1.0, 1.0, numpy.log10(1.0)]
self.assertEqual(y_pred, pipeline.predict(X).tolist())
self.assertEqual([y_predt for i in range(0, 3)], pipeline.predict_transform(X).tolist())
def sin(x):
x = numpy.pi*(x - 0.5)
return numpy.sin(x)
def tanh(x):
x = 2 * numpy.pi * (x - 0.5)
return numpy.tanh(x)
train = numpy.linspace(0, 1, 9)
test = numpy.linspace(0.01, 0.99, 12)
X = DataFrame(test.reshape(-1, 1), columns = ["x1"])
store_csv(X, "BSpline")
for fun in [gaussian, sin, tanh]:
name = fun.__name__.capitalize() + "BSpline"
bspline = make_interp_spline(train, fun(train), k = 3)
regressor = LinearRegression()
regressor.coef_ = numpy.array([1])
regressor.intercept_ = numpy.array([0])
pipeline = PMMLPipeline([
("regressor", regressor)
], predict_transformer = BSplineTransformer(bspline))
store_pkl(pipeline, name)
y = DataFrame(pipeline.predict_transform(X), columns = ["y", "bspline(predict(y))"])
store_csv(y, name)
