def plot_error_hist_per_model():
fig = plt.figure()
for (model, sub_idx) in zip(["*esn*", "*arima_r*", "*arima_ind*", "*wavelet*", "*CBR*"],
(1, 2, 3, 4, 5)):
ax = fig.add_subplot(3, 2, sub_idx, projection='3d')
wildcards = ["_run_[0-9]*_bc-data.txt"]
model_path = os.path.join(SG_SIM_PATH, "isgt-env-replace-3-of-7", model)
matrix = _get_errors_as_dataframe(model_path, *wildcards).as_matrix()
#bins = np.array(range(0, 11000, 100) + [35000])
bins = np.array(range(0, 2100, 100))
colors = [cm.flag(i) for i in np.linspace(0, 1, len(bins-1))]
for pred in range(matrix.shape[1] - 1, -1, -1):
(hist, _) = np.histogram(matrix[:,pred], bins=bins)
x = np.array(range(len(bins[:-1])))
x = np.array(bins[:-1])
ax.bar(x, hist, pred, zdir='y', color=colors, edgecolor=colors)
#ax.set_xticks(x + 0.5)
#ax.set_xticklabels([str(bin) for bin in bins[:-1]])
ax.set_xlabel('Error with model ' + model)
ax.set_ylabel('Predictor number')
ax.set_zlabel('Number of days with error')
def plot_watersheds_3d(nodes_in_watersheds, landscape, ds):
"""
Plot all or some watersheds in the landscape using different colors for different watersheds in 3D. Using the
standard method.
:param nodes_in_watersheds: List of arrays. Each array have all indices in the watershed.
:param landscape: Landscape object with all data.
:param ds: Downsampling factor for only plotting every ds point.
:return: Plot watersheds in 3D using the standard method.
"""
# Construct the (x, y)-coordinate system
x_grid = np.linspace(landscape.x_min, landscape.x_max, landscape.num_of_nodes_x)
y_grid = np.linspace(landscape.y_max, landscape.y_min, landscape.num_of_nodes_y)
x_landscape, y_landscape = np.meshgrid(x_grid[0::ds], y_grid[0::ds])
z_landscape = landscape.arr[0::ds, 0::ds]
# Plot the landscape in 3D
fig, ax = plt.subplots(subplot_kw=dict(projection='3d'))
ax.plot_surface(x_landscape, y_landscape, z_landscape, cmap=plt.get_cmap('terrain'), zorder=0)
# Plot all watersheds with over n nodes
large_watersheds = ([watershed for watershed in nodes_in_watersheds
if len(watershed) > 5000])
nr_of_large_watersheds = len(large_watersheds)
colors = iter(cm.flag(np.linspace(0, 1, nr_of_large_watersheds)))
# Plot all watersheds on top of the terrain
for i in range(len(large_watersheds)):
row_col = util.get_row_and_col_from_indices(large_watersheds[i], landscape.num_of_nodes_x)
x = landscape.x_min + row_col[0::ds, 1] * landscape.step_size
y = landscape.y_max - row_col[0::ds, 0] * landscape.step_size
z = landscape.arr[row_col[0::ds, 0], row_col[0::ds, 1]]
ax.scatter(x, y, z, c=next(colors), s=30, lw=0, zorder=1)
plt.show()
targetx = 0.7800808571829128
targety = 0.2944330689467324
# for thick, realistic cups
targetx = 0.7279635069051532
targety = 0.403791382614881
for zoom in xrange(110):
tfile = '/media/scratch/cuptemp_wada'
N = 1024
width = size0 * 0.8**zoom
print zoom, width, targetx - width/2, targety - width/2
args = ['./cupgame-nowait-mapgen', N, 2.0, 1.00,
targetx - width/2, targetx + width/2,
targety - width/2, targety + width/2, tfile]
check_call([str(a) for a in args])
tt = np.fromfile(tfile, dtype='int32').reshape(N,N)
# for thin cups
#out = cm.flag(Normalize(vmin=-3000, vmax=3000)(tt))
# for thick cups
out = cm.flag(Normalize(vmin=-20, vmax=20)(tt))
out[tt == 0] = [1.0,1.0,1.0,1.0]
imsave(join(outdir, 'wada_%04d.png' % zoom), out)
#!/usr/bin/env python
# coding: utf-8
# In[2]:
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import numpy as np
fig, (ax1, ax2, ax3, ax4, ax5, ax6, ax7) = plt.subplots(nrows=7, sharex=True)
x = np.linspace(0, 2 * np.pi, 100)
for i in range(30):
y = i * np.sin(x)
ax1.plot(x, y, color=cm.rainbow(i / 30.0))
ax2.plot(x, y, color=cm.Reds(i / 30.0))
ax3.plot(x, y, color=cm.binary(i / 30.0))
ax4.plot(x, y, color=cm.PiYG(i / 30.0))
ax5.plot(x, y, color=cm.twilight(i / 30.0))
ax6.plot(x, y, color=cm.Pastel1(i / 30.0))
ax7.plot(x, y, color=cm.flag(i / 30.0))
ax1.set_xlim(0, 2 * np.pi)
fig.show()
# In[ ]:
