def get_heatmap_from_polar_all_predictors(predictions, curve_secant, intersection_angle):
# Set up the grid to sample the RBF at
r_min = 2.0
r_max = 25.0
r_resolution = 100
phi_min = 0.0
phi_max = np.pi
phi_resolution = 100
R = np.rot90(np.tile(np.linspace(r_min, r_max, r_resolution), (phi_resolution, 1)))
Phi = np.tile(np.linspace(phi_min, phi_max, phi_resolution), (r_resolution, 1))
D = np.zeros((np.shape(Phi)[0]-1, np.shape(Phi)[1]-1))
# Set up the RBF Interpolator
angle_steps = np.linspace(0., np.pi, SAMPLE_RESOLUTION)
rbfPhi = []
rbfR = []
for pred in predictions:
rbfPhi.extend(list(angle_steps))
rbfR.extend(list(pred))
# Scale factor to balance the coordinate influence in determination of r for rbf
scale_phi = 0.5*(r_min + r_max) #1/(phi_max - phi_min)
scale_r = 1#/(r_max - r_min)
rbfi = RBFValley(np.array(rbfPhi)*scale_phi, np.array(rbfR)*scale_r, 0.7)
# Sample the RBF
for j in range(np.shape(Phi)[0]-1):
for k in range(np.shape(Phi)[1]-1):
D[j,k] = rbfi(scale_phi*Phi[j,k], scale_r*R[j,k])
# Transform RBF grid into XY-Space for heatmap
X, Y = get_cartesian_from_polar(R, Phi, curve_secant, intersection_angle)
return X, Y, D
def plot_polar_probability_heatmap(predicted_proba, curve_secant, intersection_angle):
"""Plot a heatmap in polar coordinates"""
prediction = np.flipud(np.rot90(predicted_proba['predictions_proba']))
bin_num = np.shape(prediction)[0]
max_radius = predicted_proba['max_radius']
min_radius = predicted_proba['min_radius']
bin_width = np.pi/(np.shape(prediction)[1])
R = np.rot90(np.tile(np.linspace(max_radius, min_radius, bin_num + 1), (np.shape(prediction)[1] + 1, 1)))
Phi = np.tile((np.linspace(0.-bin_width/2, np.pi+bin_width/2, np.shape(prediction)[1] + 1) - bin_width/2), (bin_num + 1, 1))
X, Y = get_cartesian_from_polar(R, Phi, curve_secant, intersection_angle)
# if rotation:
# rot_phi, rot_c = rotation
# target_shape = np.shape(X)
# coords = np.transpose(np.vstack((X.ravel(), Y.ravel())))
# rot_coords = np.transpose(rotate_xy(coords, rot_phi, rot_c))
# X = np.reshape(rot_coords[0], target_shape)
# Y = np.reshape(rot_coords[1], target_shape)
ax = plt.gca()
p = ax.pcolormesh(X, Y, prediction, cmap="Oranges")
plt.gcf().colorbar(p)
