def show_result(poly_data: pandas.DataFrame,
regression_model: RegressionResultsWrapper,
predict_model: pandas.DataFrame, y_param: str, degree: int):
poly_features = [
i for i in poly_data.columns.values if i.startswith('power_')
]
# #7
pred_model = regression_model.predict(
sm.add_constant(predict_model[poly_features]))
plot_polynomial(poly_data, pred_model, y_param)
# #9
print('-----coefficient of degree {deg}------'.format(deg=degree))
print_coefficient(regression_model.params)
def expression_fields(
xs: np.ndarray,
ys: np.ndarray,
results: regres,
n_ticks: int = 400,
) -> Tuple[np.ndarray, np.ndarray, Tuple[int, int]]:
mx = np.max((xs[:, 1]))
mn = np.min(xs[:, 1])
xx = np.linspace(mn, mx, n_ticks)
mx = np.max((xs[:, 2]))
mn = np.min(xs[:, 2])
yy = np.linspace(mn, mx, n_ticks)
X, Y = np.meshgrid(xx, yy)
shape = X.shape
Xf = X.flatten()
Yf = Y.flatten()
XY = np.hstack((np.ones(
(Xf.shape[0], 1)), Xf[:, np.newaxis], Yf[:, np.newaxis]))
Z = results.predict(XY)
return (XY[:, 1::], Z, shape)
def get_rss(model: RegressionResultsWrapper, data: list, input_model: list,
param_name: str) -> float:
prediction = model.predict(sm.add_constant(data[input_model]))
residuals = data[param_name] - prediction
rss = (residuals**2).sum()
return rss