class PolyFeaturesImplementation(EncodedInvariantImplementation):
""" Class for application of PolynomialFeatures operation on data,
where only not encoded features (were not converted from categorical using
OneHot encoding) are used
:param params: optional, dictionary with the arguments
"""
def __init__(self, **params: Optional[dict]):
super().__init__()
if not params:
# Default parameters
self.operation = PolynomialFeatures(include_bias=False)
else:
# Checking the appropriate params are using or not
poly_params = {k: params[k] for k in
['degree', 'interaction_only']}
self.operation = PolynomialFeatures(include_bias=False,
**poly_params)
self.params = params
def get_params(self):
return self.operation.get_params()
def get_params(self, deep=True):
return PolynomialFeatures.get_params(self, deep=deep)
plt.plot(data_x, curve_polynomial, 'b.')
plt.xlabel("np.linspace(0, 8, 32)")
plt.ylabel("curve fitting using Linear")
hypo = np.dot(data_x, linear_reg.coef_.T) + linear_reg.intercept_
hypo_fail = np.dot(data_x, linear_reg_fail.coef_.T) + linear_reg_fail.intercept_
print(mean_squared_error(hypo, curve_linear)) # MSE: 0.00069
print(mean_squared_error(hypo_fail, curve_polynomial)) # MSE: 6.40752
elif method_reg == "non-linear":
method_reg = value
elif method_reg == "polynomial":
poly_features_1 = PolynomialFeatures(degree = 3)
linear_reg = LinearRegression()
linear_reg.fit(poly_features_1.fit_transform(data_x), curve_linear)
print(poly_features_1.get_params())
fit_x = np.linspace(start = 0, stop = sample_cnt/4, num = 1024).reshape(-1, 1)
fit_linear = linear_reg.predict(poly_features_1.fit_transform(fit_x))
plt.subplot(2, 2, 3)
plt.plot(fit_x, fit_linear, 'r-')
plt.plot(data_x, curve_linear, 'b.')
plt.xlabel("np.linspace(0, 8, 32)")
plt.ylabel("curve fitting using Polynomial")
poly_features_3 = PolynomialFeatures(degree = 3)
linear_reg_best = LinearRegression()
linear_reg_best.fit(poly_features_3.fit_transform(data_x), curve_polynomial)
fit_x = np.linspace(start = 0, stop = sample_cnt/4, num = 1024).reshape(-1, 1)
fit_linear_best = linear_reg_best.predict(poly_features_3.fit_transform(fit_x))
