def plotWindow(wavelength, signal):
"""
Plots the loaded spectra for visual inspection.
"""
pyplot.figure("Loaded spectra")
pyplot.plot(wavelength, signal)
pyplot.show()
def plotArray(self, arraylist):
print("Plot Array Called")
tmp = self.arrayData[arraylist.currentRow()]
plt.figure(self.arrayName[arraylist.currentRow()])
for x in range(tmp.columns.size):
# add mask to avoid nan data
mask = np.isfinite(tmp.iloc[:, x].values)
plt.plot(tmp.index[mask], tmp.iloc[:, x].values[mask], "b-")
plt.show()
def plotCSV(self, csvlist):
# do plot csv figure
# After right click on the menu
tmp = self.csvData[csvlist.currentRow()]
plt.figure(self.csvName[csvlist.currentRow()])
# add mask to avoid nan
for x in range(tmp.columns.size):
if x == 0:
mask = np.isfinite(tmp.iloc[:, x].values)
print(mask)
plt.plot(tmp.index[mask], tmp.iloc[:, x].values[mask], "b-")
elif x == 1:
mask = np.isfinite(tmp.iloc[:, x].values)
print(mask)
plt.plot(tmp.index[mask], tmp.iloc[:, x].values[mask], "r-")
else:
plt.plot(tmp.index, tmp.iloc[:, x].values, "g-")
plt.show()
def varexp(line):
"""
Spyder's variable explorer magic
Used to generate plots, histograms and images of the variables displayed
on it.
"""
ip = get_ipython() #analysis:ignore
funcname, name = line.split()
try:
import guiqwt.pyplot as pyplot
except:
import matplotlib.pyplot as pyplot
__fig__ = pyplot.figure();
__items__ = getattr(pyplot, funcname[2:])(ip.user_ns[name])
pyplot.show()
del __fig__, __items__
def MNIST_Plot_Image_Pairs(Img_Orig, Img_Recon, **kwargs):
# This function plots pairs of MNIST images from two matrices in a grid
# by reshaping 784x1 column vectors from supplied matrices into pairs of
# two 28x28 images concatenated side by side.
# Inputs
# ============================================================
# Img_Orig : 784 x N_Plots First image
# Img_Recon: 784 x N_Plots Second image
# Optional inputs
# N_Horiz : integer Number of pairs of images on horizonal axis
# N_Vert : integer Number of pairs of images on vertical axis
# str : string Figure title
#
# Output
# ============================================================
# A figure with pairs of images
if Img_Orig.shape != Img_Recon.shape:
print('Error: Dimension mismatch!')
if Img_Orig.shape[0] != 784:
print('Error: Weight matrix does not have 784 rows!')
if "str" in kwargs:
str = kwargs["str"]
else:
str = 'MNIST: Pairs of images'
# Number of images to reconstruct and plot
N_Plots = Img_Orig.shape[1]
if ("N_Horiz" in kwargs) and ("N_Vert" in kwargs):
N_Horiz = kwargs["N_Horiz"]
N_Vert = kwargs["N_Vert"]
if N_Horiz * N_Vert == N_Plots:
Do_Init = False
else:
Do_Init = True
else:
Do_Init = True
if Do_Init == True:
N_Horiz = int(np.floor(np.sqrt(N_Plots)))
N_Vert = int(np.ceil(np.sqrt(N_Plots)))
if N_Horiz * N_Vert < N_Plots:
N_Horiz = N_Horiz + 1
# Vertical padding
V_Pad = 2 # pixels
# Horizontal padding
H_Pad = 2 # pixels
# Number of pixels in width (height as well, this is a square image)
N = 28 # pixels
# Allocate memory for the combined image
A = np.zeros([
N * N_Horiz + H_Pad * (N_Horiz - 1),
2 * N * N_Vert + V_Pad * (N_Vert - 1)
])
# Counter for the image sample to be plotted
cnt = 0
for i in range(N_Horiz):
for j in range(N_Vert):
# Calculating image position on the grid
idx_H1 = i * (N + H_Pad)
idx_H2 = (i + 1) * (N + H_Pad) - H_Pad
idx_V1 = j * (2 * N + V_Pad)
idx_V2 = (j + 1) * (2 * N + V_Pad) - V_Pad
# Reshape a 784x1 vector into a 28x28 image. Do the same operation for the reconstructed image
A[idx_H1:idx_H2, idx_V1:idx_V2] = np.concatenate([
Img_Orig[:, cnt].reshape(N, N, order='F'),
Img_Recon[:, cnt].reshape(N, N, order='F')
],
axis=1)
cnt = cnt + 1
if cnt == N_Plots:
break
plt.figure(str)
plt.gray() # colormap
plt.imshow(A)
plt.show()
