def run(X, Y):
N = len(Y)
# Z possui os valores obtidos com o somatório
# [ [real, imaginario], [real, imaginario], ... ]
Z = numpy.empty(shape=[N, 2])
# mag é o array com os módulos
mag = numpy.empty(N)
# angle é o array com os angulos
angle = numpy.empty(N)
for k in range(0, N):
for n in range(0, N):
b = 2*math.pi*k*n / N
Z[k][0] += Y[n]*math.cos(-b)
Z[k][1] += Y[n]*math.sin(-b)
mag[k] = math.sqrt(math.pow(Z[k][0], 2) + math.pow(Z[k][1], 2))
# plotar espectro de linhas com mag no eixo Y
plt.stem(X, mag)
plt.title('Espectro de Linhas')
plt.ylabel('Magnitude')
print(Z)
print(mag)
print(angle)
plt.show()
def plot(self,
typ='s3',
title='',
xl=False,
yl=False,
log=False,
stem=True,
color='b'):
"""
"""
if typ == 's3':
indices = self.ind3
tl = indices[:, 0]
C = []
for l in np.unique(tl):
k = np.where(tl == l)
a = np.real(np.sum(self.s3[:, k] * np.conj(self.s3[:, k])))
C.append(a)
C = np.real(np.array(C))
Cs = np.sqrt(C)
if log:
Cs = 20 * log10(Cs)
if stem:
plt.stem(np.unique(tl), Cs, markerfmt=color + 'o')
else:
plt.plot(np.unique(tl), Cs, color=color)
#plt.axis([0,max(tl),0,5])
plt.title(title)
if xl:
plt.xlabel('degree l')
if yl:
plt.ylabel('Integrated Module of coeff')
def Histogram_Equalization(image):
# image1=[[1,2,3],[2,1,3],[1,3,3]]
min_value = np.min(image)
max_value = np.max(image)
j, k = np.shape(image)
frequency = []
pdf = []
cdf = []
count = 0
total_pixel = 0
sum1 = 0
for i in range(min_value, max_value + 1):
for a in range(j):
for b in range(k):
if image[a][b] == i:
print(i)
count = count + 1
print(count)
frequency.append((i, count))
total_pixel = total_pixel + count
count = 0
for intensity, no_of_repeat in frequency:
pdf.append(no_of_repeat / total_pixel)
for value in pdf:
sum1 = sum1 + value
cdf.append(sum1)
plt.stem(cdf)
plt.show()
def histogram_plot():
global image, extension, image_count, log_path, max_image_count, view_value, plot_number
if (image.max() != 0):
if (
plot_number == 121
): #because this function will be called two times in display_original()
fig.clf()
fig.add_subplot(plot_number) #dynamically changing the grid plot
im = plt.imshow(image, vmin=0, vmax=255)
plt.set_cmap('gray')
image_histogram = np.zeros(256) #contains histogram of image
#create the image histogram
for i in range(image.shape[0]):
for j in range(image.shape[1]):
intensity_value = int(
abs(image[i][j])
) #abs() is taken in case of negative image intensities
image_histogram[
intensity_value] = image_histogram[intensity_value] + 1
fig.add_subplot(plot_number + 1)
plt.stem(np.arange(256), image_histogram,
markerfmt=' ') #plots a stem image of histogram
canvas.draw()
view_value = "100" #set the callback so that the display image option is enabled
var2.set(view_value)
else:
messagebox.showerror("Error", "Please load an image first.")
def plot(self, typ="s3", title="", xl=False, yl=False, log=False, stem=True, color="b"):
"""
"""
if typ == "s3":
indices = self.ind3
tl = indices[:, 0]
C = []
for l in np.unique(tl):
k = np.where(tl == l)
a = np.real(np.sum(self.s3[:, k] * np.conj(self.s3[:, k])))
C.append(a)
C = np.real(np.array(C))
Cs = np.sqrt(C)
if log:
Cs = 20 * log10(Cs)
if stem:
plt.stem(np.unique(tl), Cs, markerfmt=color + "o")
else:
plt.plot(np.unique(tl), Cs, color=color)
# plt.axis([0,max(tl),0,5])
plt.title(title)
if xl:
plt.xlabel("degree l")
if yl:
plt.ylabel("Integrated Module of coeff")
def demo(text=None):
from nltk.corpus import brown
from matplotlib import pylab
import jieba
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
with open('flypaper_short.txt', 'r') as file:
comments = file.read()
# segment the word
seg_list = jieba.cut(comments)
text = (" ".join(seg_list))
#print(text)
pattern = re.compile("\n")
matches = pattern.finditer(text)
#for match in matches:
#print(match.start())
MIN_PARAGRAPH = 100 # minimum length of a paragraph
last_break = 0
pbreaks = [0]
for pb in matches:
if pb.start() - last_break < MIN_PARAGRAPH:
continue
else:
pbreaks.append(pb.start())
last_break = pb.start()
print(pbreaks)
pb_iter = pbreaks.__iter__()
current_par_break = next(pb_iter)
if current_par_break == 0:
try:
current_par_break = next(pb_iter) # skip break at 0
except StopIteration:
raise ValueError(
"No paragraph breaks were found(text too short perhaps?)")
tt = TextTilingTokenizer(w=10, k=3, demo_mode=True)
#if text is None: text = brown.raw()[:10000]
s, ss, d, b = tt.tokenize(text)
print(b)
pylab.xlabel("Sentence Gap index")
pylab.ylabel("Gap Scores")
pylab.plot(range(len(s)), s, label="Gap Scores")
pylab.plot(range(len(ss)), ss, label="Smoothed Gap scores")
pylab.plot(range(len(d)), d, label="Depth scores")
pylab.stem(range(len(b)), b)
pylab.legend()
pylab.show()
def plot_feature_class_corr_matrix(df, labels, cols):
corr = []
for i in xrange(0, df.shape[1]):
corr.append(stats.pointbiserialr(labels, df.icol(i)))
pos = np.arange(1, df.shape[1] + 1)
c, p = zip(*corr)
plt.figure(figsize=(10, 10))
plt.grid()
plt.stem(pos, c)
plt.xticks(pos, cols, rotation=90, fontsize=8)
plt.ylabel('Correlation')
plt.title('Point biserial Correlation - Features v. Class')
plt.savefig('Graphs/BiSerialCorr.png', bbox_inches='tight')
def demo(text=None):
from nltk.corpus import brown
from matplotlib import pylab
tt = TextTilingTokenizer(demo_mode=True)
if text is None: text = brown.raw()[:10000]
s, ss, d, b = tt.tokenize(text)
pylab.xlabel("Sentence Gap index")
pylab.ylabel("Gap Scores")
pylab.plot(range(len(s)), s, label="Gap Scores")
pylab.plot(range(len(ss)), ss, label="Smoothed Gap scores")
pylab.plot(range(len(d)), d, label="Depth scores")
pylab.stem(range(len(b)), b)
pylab.legend()
pylab.show()
def demo(text=None):
from nltk.corpus import brown
from matplotlib import pylab
tt = TextTilingTokenizer(demo_mode=True)
if text is None: text = brown.raw()[:10000]
s, ss, d, b = tt.tokenize(text)
pylab.xlabel("Sentence Gap index")
pylab.ylabel("Gap Scores")
pylab.plot(range(len(s)), s, label="Gap Scores")
pylab.plot(range(len(ss)), ss, label="Smoothed Gap scores")
pylab.plot(range(len(d)), d, label="Depth scores")
pylab.stem(range(len(b)), b)
pylab.legend()
pylab.show()
def InitialFinalSilenceRemoved(sig):
# Removes beginning and end silence periods of a wavfile
# Input: sig, i.e. wavfile
# Output: new_sig, i.e. wavfile without beginning and end silence periods
#########################################################################
window = 512
hop = window / 2
energy = []
i = 0
energy_index = []
while i < (len(sig) - window):
chunk = sig[i:i + window][np.newaxis]
energy.append(chunk.dot(chunk.T)[0][0])
energy_index.append(i)
i = i + hop
energy = np.array(energy)
energy_thresh = 0.1 * np.mean(energy)
significant_indices = np.where(energy > energy_thresh)[0]
if significant_indices[0] == 0:
start_point_sample = 0
else:
start_point_sample = (significant_indices[0] - 1) * hop
if significant_indices[-1] == len(energy) - 1:
end_point_sample = len(energy) * hop
else:
end_point_sample = (significant_indices[-1] + 1) * hop
new_sig = sig[start_point_sample:end_point_sample + 1]
if plot:
plt.figure('figure from InitialFinalSilenceRemoved')
plt.subplot(3, 1, 1)
plt.plot(range(len(sig)), sig)
plt.ylabel('amplitude')
plt.title('Remove initial and final silences')
plt.subplot(3, 1, 2)
plt.plot(energy_index, energy)
plt.ylabel('energy')
plt.stem([start_point_sample, end_point_sample], [5, 5], 'k')
plt.subplot(3, 1, 3)
plt.plot(new_sig)
plt.ylabel('amplitude')
plt.xlabel('sample number')
plt.show()
return new_sig
def hist_mtx(mtx, tstr=''):
"""
Given a piano-roll matrix, 128 MIDI piches x beats, plot the pitch class histogram
"""
i_min, i_max = np.where(mtx.mean(1))[0][[0, -1]]
P.figure(figsize=(14.5, 8))
P.stem(np.arange(i_max + 1 - i_min), mtx[i_min:i_max + 1, :].sum(1))
ttl = 'Note Frequency'
if tstr: ttl += ': ' + tstr
P.title(ttl, fontsize=16)
t = P.xticks(np.arange(0, i_max + 1 - i_min, 3),
pc_labels[i_min:i_max + 1:3],
fontsize=14)
P.xlabel('Pitch Class', fontsize=14)
P.ylabel('Frequency', fontsize=14)
ax = P.axis()
P.axis(xmin=-0.5)
P.grid()
def PlotProcessFragmentation(title, data, output):
"""Plots the Fragmentation vs size for a single process.
Args:
title: Title of the graph
data: Data to plot. Should contain 'size' and 'fragmentation' entries.
output: Filename to save the result to.
"""
plt.figure(figsize=(16, 8))
plt.title(title)
plt.stem(data['data']['size'], data['data']['fragmentation'])
plt.xscale('log', base=2)
plt.yscale('linear')
plt.ylim(ymin=0, ymax=100)
plt.xlabel('Size (log)')
plt.ylabel('Fragmentation (%)')
plt.savefig(output, bbox_inches='tight')
plt.close()
def plot_normalized_frequency_distribution(n_fqdist, n=20):
text_type = unicode
pylab.grid(True, color="silver")
pylab.title(
'Normalized Word Frequency Distribution. TOP {n} Samples'.format(n=n))
freqs = n_fqdist.values()[0:n + 1]
samples = n_fqdist.keys()[0:n + 1]
pylab.stem(freqs)
pylab.xticks(range(len(samples)), [text_type(s) for s in samples],
rotation=90)
pylab.xlabel("Samples")
pylab.ylabel("Normalized Word Frequency")
pylab.show()
return True
def _PlotProcess(all_data: dict, pid: int, output_prefix: str):
"""Represents the allocation size distribution.
Args:
all_data: As returned by _ParseTrace().
pid: PID to plot the data for.
output_prefix: Prefix of the output file.
"""
data = all_data[pid]
logging.info('Plotting data for PID %d' % pid)
# Allocations vs size.
plt.figure(figsize=(16, 8))
plt.title('Allocation count vs Size - %s - %s' %
(data['name'], data['labels']))
plt.xscale('log', base=2)
plt.yscale('log', base=10)
plt.stem(data['data']['size'], data['data']['count'])
plt.xlabel('Size (log)')
plt.ylabel('Allocations (log)')
plt.savefig('%s_%d_count.png' % (output_prefix, pid), bbox_inches='tight')
plt.close()
# CDF.
plt.figure(figsize=(16, 8))
plt.title('CDF of allocation size - %s - %s' % (data['name'], data['labels']))
cdf = np.cumsum(100. * data['data']['count']) / np.sum(data['data']['count'])
for value in [512, 1024, 2048, 4096, 8192]:
index = np.where(data['data']['size'] == value)[0]
cdf_value = cdf[index]
plt.axvline(x=value, ymin=0, ymax=cdf_value / 100., color='lightgrey')
plt.step(data['data']['size'], cdf, color='black', where='post')
plt.ylim(ymin=0, ymax=100)
plt.xlim(xmin=10, xmax=1e6)
plt.xscale('log', base=2)
plt.xlabel('Size (log)')
plt.ylabel('CDF (%)')
plt.savefig('%s_%d_cdf.png' % (output_prefix, pid),
bbox_inches='tight',
dpi=300)
plt.close()
def PlotProcessWaste(title, data, output):
"""Plots the Unused memory vs size for a single process.
Args:
title: Title of the graph
data: Data to plot. Should contain 'size' and 'unused' entries.
output: Filename to save the result to.
"""
plt.figure(figsize=(16, 8))
plt.title(title)
plt.xscale('log', base=2)
plt.yscale('log', base=2)
plt.stem(data['data']['size'][data['data']['unused'] != 0],
data['data']['unused'][data['data']['unused'] != 0])
plt.ylim(ymin=1, ymax=2**20)
plt.xlabel('Size (log)')
plt.ylabel('Unused Size (log)')
plt.savefig(output, bbox_inches='tight')
plt.close()
def define_params(ts_diff):
lag_acf = acf(ts_diff, nlags=20)
lag_pacf = pacf(ts_diff, nlags=20, method='ols')
# q的获取:ACF图中曲线第一次穿过上置信区间.这里q取2
plt.subplot(121)
plt.stem(lag_acf)
plt.axhline(y=0, linestyle='--', color='gray')
plt.axhline(y=-1.96 / np.sqrt(len(ts_diff)), linestyle='--', color='gray') # lowwer置信区间
plt.axhline(y=1.96 / np.sqrt(len(ts_diff)), linestyle='--', color='gray') # upper置信区间
plt.title('Autocorrelation Function')
# p的获取:PACF图中曲线第一次穿过上置信区间.这里p取2
plt.subplot(122)
plt.stem(lag_pacf)
plt.axhline(y=0, linestyle='--', color='gray')
plt.axhline(y=-1.96 / np.sqrt(len(ts_diff)), linestyle='--', color='gray')
plt.axhline(y=1.96 / np.sqrt(len(ts_diff)), linestyle='--', color='gray')
plt.title('Partial Autocorrelation Function')
plt.tight_layout()
plt.show()
def main():
n1 = int(input('Enter the number of trials with a large p: '))
p1 = float(input('Enter the probability of trial: '))
'''Bernoulli Random Variable'''
xbe = [0, 1]
ybe = [1-p1, p1]
plt.title("Bernoulli Distribution Stem Plot")
plt.stem(xbe, ybe, '-.')
plt.xlabel("X : Bernoulli Random Variable")
plt.ylabel("PMF")
plt.show()
'''Binomial Random Variable Plot'''
xbi, ybi = plot_Binomial(trial = n1, probability = p1)
plt.subplot(211)
plt.title("Binomial Distribution Stem Plot")
plt.stem(xbi, ybi, '-.')
plt.xlabel("X : Binomial Random Variable")
plt.ylabel("PMF")
'''Geometric Random Variable Plot'''
xge, yge = plot_Geometric(success = n1, probability = p1)
plt.subplot(212)
plt.title("Geometric Distribution Stem Plot")
plt.stem(xge, yge, '-.')
plt.xlabel("X : Geometric Random Variable")
plt.ylabel("PMF")
plt.show()
n2 = int(input('Enter the number of trials with less p.: '))
p2 = float(input('Enter the probability of trial: '))
t = int(input('Enter how long the incident took place.: '))
'''Poisson Random Variable Plot'''
xpo, ypo = plot_Poisson(trial = n2, probability = p2)
plt.subplot(211)
plt.title("Poisson Distribution Stem Plot")
plt.xlabel("X : Poisson Random Variable")
plt.ylabel("PMF")
plt.stem(xpo, ypo, '-.')
'''Exponential Random Variable Plot'''
l = n2 * p2
xex, yex = plot_Exponential(lamb = l, time = t)
plt.subplot(212)
plt.title("Exponential Distribution Plot")
plt.xlabel("X : Exponential Random Variable")
plt.ylabel("PDF")
plt.plot(xex, yex)
plt.show()
def _process(self, data):
num = math.ceil(len(data) / 2)
fftf = np.fft.fft(data) * 2 / len(data)
if not self.parameter['mph']:
self.parameter['mph'] = np.mean(data) / 3
frequency = process.detect_peaks(abs(fftf)[0:num],
mph=self.parameter['mph'],
mpd=self.parameter['mpd'],
threshold=self.parameter['thre'])
if self.parameter['show']:
plt.figure()
for i in frequency:
plt.scatter(i - 1, abs(fftf)[0:num][i], color='r')
plt.stem(abs(fftf)[1:num], color='#87CEEB')
plt.show()
return [i for i in frequency / 12 if i < self.parameter['mc']]
def stem_timeseries_multi(timeseries_array: List[Timeseries], title, xlabel, ylabel, separate):
matplotlib.style.use('default')
fig = plt.figure()
bottom = None
for ts in timeseries_array:
for y in ts.y:
if bottom is None or y < bottom:
bottom = y
for ts in timeseries_array:
plt.stem(ts.x, ts.y, label=ts.label, bottom=bottom)
set_disp(title, xlabel, ylabel)
plt.legend()
fig = plt.gcf()
plt.show()
return fig, mpld3.fig_to_html(fig)
def demo(text=None):
'''
use the bounary together with the pseudo sentences to evaluate the quality of segmentation.
:param text:
:return:
'''
from nltk.corpus import brown
from matplotlib import pylab
tt = TextTilingTokenizer(w=40, k=20, demo_mode=True)
with open('flypaper_short.txt', 'r') as file:
text = file.read()
if text is None: text = brown.raw()[:10000]
s, ss, d, b = tt.tokenize(text)
print(b)
pylab.xlabel("Sentence Gap index")
pylab.ylabel("Gap Scores")
pylab.plot(range(len(s)), s, label="Gap Scores")
pylab.plot(range(len(ss)), ss, label="Smoothed Gap scores")
pylab.plot(range(len(d)), d, label="Depth scores")
pylab.stem(range(len(b)), b)
pylab.legend()
pylab.show()
def myImHist(image):
#image1=[[1,2,3],[2,1,3],[1,3,3]]
min_value = np.min(image)
max_value = np.max(image)
j, k = np.shape(image)
frequency = []
pdf = []
count = 0
total_pixel = 0
for i in range(min_value, max_value + 1):
for a in range(j):
for b in range(k):
if image[a][b] == i:
print(i)
count = count + 1
print(count)
frequency.append((i, count))
total_pixel = total_pixel + count
count = 0
for intensity, no_of_repeat in frequency:
pdf.append(no_of_repeat / total_pixel)
plt.stem(pdf)
plt.show()
def zpFFTsizeExpt(): f = 110.0 fs = 1000.0 t = np.arange(0,1,1.0/fs) x = cos(2 * pi * f * t) xseg = x[0:256] w1 = np.hamming(256) w2 = np.hamming(512) X1 = fft(xseg * w1) X2 = fft(x[0:512] * w2) X3 = fft(xseg * w1, 512) mx1 = abs(X1) mx2 = abs(X2) mx3 = abs(X3) fx1 = fs * np.arange(256) / 256 fx2 = fs * np.arange(512) / 512 plt.xlim(0,150) plt.stem(fx1[0:80],mx1[0:80],'y') plt.stem(fx2[0:80],mx2[0:80],'r') plt.stem(fx2[0:80],mx3[0:80],'b') plt.show()
N1 = int(N / 10) N2 = int(N * 10) Ceros1 = np.zeros(N1) Ceros2 = np.zeros(N2) resultado = np.concatenate((Ceros1, signal, Ceros2), axis=None) plt.plot(resultado) plt.show() sp = np.fft.fft(resultado) plt.stem(np.absolute(sp)[0:500]) plt.show() #plt.plot(np.angle(sp)[0:500]) #plt.show() cuadrado = (np.absolute(sp)[0:500])**2 energia = integrate.simps(cuadrado) cuadrado0 = (np.absolute(sp)[f0])**2 plt.plot(20 * np.log10(np.absolute(sp)[0:500])) plt.show() asd = max(cuadrado)
###########################################################
# nlags: number of samples for autocorrelation to be returned for
# autucorrelation
lag_acf = acf(training_log_diff, nlags=30)
# partial autocorrelation
lag_pacf = pacf(training_log_diff, nlags=30, method='ols')
plt.figure(figsize=(12, 5))
plt.xlabel("no. of lag")
plt.ylabel("lag")
plt.title("ACF PLOT")
plt.axhline(y=0, linestyle='-', color='black')
plt.axhline(y=1.96 / np.sqrt(len(training)), linestyle='--', color='gray')
plt.axhline(y=-1.96 / np.sqrt(len(training)), linestyle='--', color='gray')
#plt.plot(lag_acf)
plt.stem(lag_acf)
# suggests seasonality period=12, hence S=12
# at S=12, lag is positive so P=1 and Q=0
# significat lag at 1 so q=1
plt.figure(figsize=(12, 5))
plt.xlabel("no. of lag")
plt.ylabel("lag")
plt.title("PACF PLOT")
plt.axhline(y=0, linestyle='-', color='black')
plt.axhline(y=1.96 / np.sqrt(len(training)), linestyle='--', color='gray')
plt.axhline(y=-1.96 / np.sqrt(len(training)), linestyle='--', color='gray')
#plt.plot(lag_acf)
plt.stem(lag_pacf)
# significant lag at 1 so p=1
s = 10
m = 40
D = random_dict(m, n)
k = 0
while 1:
x = get_sparse_x(n, s)
y = np.dot(D, x)
x_naive = omp_naive(D, y)
x_scikit = orthogonal_mp(D, y, s)
error_naive = norm(x - x_naive.reshape(n, 1))
error_scikit = norm(x - x_scikit.reshape(n, 1))
k += 1
if error_naive < 1e-3 and error_scikit > 1:
print x
break
print k
import matplotlib.pylab as plt
nr = np.arange(0, n)
plt.subplot(311)
plt.stem(nr, x)
plt.subplot(312)
plt.stem(nr, x_naive)
plt.subplot(313)
plt.stem(nr, x_scikit)
plt.show()
N = 1024 t = np.arange(0,1,1/N) f0 = 1 f1 = 100 sig = sig.chirp(t, f0, 1, f1, 'linear',90) SIG = fftshift(fft(sig)) F = np.linspace(-N/2,N/2,N) AVGR = sma(np.real(SIG),0.05*len(sig)) AVGI = sma(np.imag(SIG),0.05*len(sig)) plt.figure() plt.subplot(3,1,1), plt.plot(t,sig) plt.subplot(3,1,2), plt.stem(F,np.real(SIG/N)) plt.subplot(3,1,2), plt.stem(F,np.imag(SIG/N),'g') plt.subplot(3,1,3), plt.stem(F,AVGR/N) plt.subplot(3,1,3), plt.stem(F,AVGI/N,'g') ################################################################### # Frome here test values #a = np.array([2.,3.,1.,2.,2.,3.,1.,3.,1.,1.]) #smaval = sma(a,3) #smahand = np.array([5./3, 6./3, 6./3, 5./3, 7./3, 6./3, 7./3, 5./3, 5./3, 2./3]) # #for idx in range(len(smaval)): # print(smaval[idx], smahand[idx])
import numpy as np
import matplotlib as mpl
import matplotlib.pylab as plt
y = [2, 3, 1]
x = np.arange(len(y))
print(x)
xlabel = ['x label']
error = np.random.rand(len(y))
plt.title("Bar Chart")
plt.barh(x, y, alpha=0.5, xerr=error) #alpha는 투명도, xerr:에러의 허용범위
plt.show()
#스템 플롯(폭이 없는 막대차트)
x = np.linspace(0.1, 2 * np.pi, 10)
plt.title("Stem Plot")
plt.stem(x, np.cos(x), '-.')
plt.show()
#파이차트
label = ['자바', '씨', '씨++', '파이썬']
sizes = [15, 30, 45, 10]
colors = ['yellowgreen', 'gold', 'lightskyblue', 'lightcoral']
plt.title('Pie Chart')
plt.pie(sizes, labels=label, colors=colors,
startangle=90) #shadow=True로 그림자 넣어줄 수도 있음
plt.axis('equal') #
plt.show()
#히스토그램(도수분포표 막대그래프로 나타낸 것)
x = np.random.randn(1000)
plt.title("histogram")
plt.figure(figsize=(13, 4))
plt.subplot(1, 5, 1)
plt.plot(x, unif)
plt.ylim((0, 4))
plt.title('(A)')
plt.ylabel('rozdělení $p$')
plt.xlabel('p')
plt.subplot(1, 5, 2)
plt.plot(x, cent)
plt.title('(B1)')
plt.xlabel('p')
plt.ylim((0, 4))
plt.subplot(1, 5, 3)
plt.plot(x, cent2)
plt.title('(B2)')
plt.xlabel('p')
plt.ylim((0, 4))
plt.subplot(1, 5, 4)
plt.plot(x, skewed)
plt.ylim((0, 4))
plt.xlabel('p')
plt.title('(C)')
plt.subplot(1, 5, 5)
plt.stem([0.9], [4], markerfmt=None)
plt.ylim((0, 4))
plt.xlim(0, 1)
plt.xlabel('p')
plt.title('(D)')
plt.tight_layout()
plt.savefig('l1-prior-mince.jpg')
DATA_f = fftshift(fft(data_f)) # Fit
DATA_nf = fftshift(fft(data_nf)) # No fit
F = np.linspace(-N/2,N/2,N)
elif method == 'file':
[fs,data] = wavfile.read("../09 Sample 15sec.wav")#,dtype=float)
data_nf = data[2048:2048+N:]
data_nf = np.reshape(np.delete(data_nf,0, 1),len(data_nf))
DATA_nf = fftshift(fft(data_nf)) # No fit
F = np.linspace(-N/2,N/2,N) #(0,N/2,N/2+1)
if method == 'signal':
plt.figure()
plt.subplot(3,1,1), plt.plot(t,data_f,t,data_nf)
plt.title("time: 2 sines")
plt.subplot(3,1,2), plt.stem(F,np.abs(DATA_f/N)), #plt.stem(np.linspace(-N/2,N/2,N),np.imag(SIN_f/N),'g')
plt.title("Spectrum Fit - Absolut No Window")
plt.subplot(3,1,3), plt.stem(F,np.abs(DATA_nf/N)), #plt.stem(np.linspace(-N/2,N/2,N),np.imag(SIN_nf/N), 'g')
plt.title("Spectrum No Fit - Absolut No Window")
elif method == 'file':
plt.figure()
plt.subplot(2,1,1), plt.plot(t,data_nf)
plt.title("time: wav file")
plt.subplot(2,1,2), plt.stem(F,np.abs(DATA_nf/N)), #plt.stem(np.linspace(-N/2,N/2,N),np.imag(SIN_nf/N), 'g')
plt.title("Spectrum No Fit - Absolut No Window")
#[wt,x] = Window.triwind(N)
[x,whan] = Window.hanwind(N)
#[x,wcos] = Window.coswind(N)
params[0, 20:50] = 0
diag = np.random.rand(n_features)
features = np.random.multivariate_normal(np.zeros(n_features), np.diag(diag), n_samples)
# Show the condition number of the gram matrix
print("cond = %.2f" % (diag.max() / diag.min()))
linear = True
if linear == True:
residuals = np.random.randn(n_samples, 1)
labels = features.dot(params.T) + residuals
else:
labels = np.array([[float(np.random.rand() < p)] for p in logistic(features.dot(params.T))])
plt.figure(figsize=(8, 4))
plt.stem(params[0])
plt.title("True parameters", fontsize=16)
plt.show()
x_init = 1 - 2 * np.random.rand(1, n_features)
n_iter = 30
l_l1 = 0.0
l_l2 = 0.1
#f and gradient
if linear == True:
f = lambda x: least_squares(x, features, labels)
grad_f = lambda x: least_squares_grad(x, features, labels)
hess_f = lambda x: least_squares_hess(x, features, labels)
step = norm(features.T.dot(features)/ n_samples, 2)
else:
import numpy as num
a = int(input('Enter the amplitude = '))
f = int(input('Enter the frequency = '))
t = num.arange(0, 2, 0.02) # for a total of 16 samples
# %generation of an impulse signal
x1 = []
for i in range(len(t)):
x1.append(1)
x2 = a * num.sin(2 * num.pi * f * t) # %generation of sine wave
y = x1 * x2
#modulation step
#for impulse signal plot
plt.stem(t, x1)
plt.title('Impulse Signal')
plt.xlabel('Time')
plt.ylabel('Amplitude ')
plt.grid(True)
plt.show()
#for sine wave plot
plt.plot(t, x2)
plt.title('Sine Wave')
plt.xlabel('Time ')
plt.ylabel('Amplitude ')
plt.grid(True)
plt.show()
#for PAM wave plot
def _draw_levels(self, f, P, l, savefig, time_idx, Xhatlow, Xhathigh,
fP_record, X, X_next_l, verbose):
"""
Draw descriptive plots at each level of mrDMD showing
the spatial modes and spectral components recorded
and/or extracted.
Parameters
----------
f : array-like
Frequencies
P : array-like
Corresponding power
l : int
The level of mrDMD currently on
savefig : boolean
Saves figure
time_idx : array_like
The array of indices associated with current window
Xhatlow : matrix
The matrix representation of the unselected matrices
Xhathigh : matrix
The matrix representation of the selected matrices
fP_record : list
The list of (f, P) tuple pairs recorded
X : matrix
The raw data matrix
X_next_l :
The data matrix to be passed to lower levels
verbose : boolean
Explicit printing of f and P recorded
"""
self.levelSet.add(l) #only plot once per level
if verbose:
print('\n\nLevel: %d' % l)
#Frequency vs. scalings of DMD
plt.figure()
plt.rc('text', usetex=True)
plt.stem(f, P, 'k')
plt.title(r'Level: %d' % l)
title = r'SpectrumLevel:%d' % l
plt.xlabel(r'Frequency')
plt.ylabel(r'DMD scaling')
plt.show()
if savefig:
plt.savefig('%s.pdf' % title)
plt.close()
if verbose:
print('Power recorded [frequency, Power]:')
print(fP_record)
print('Added to time_idx %d - %d' % (time_idx[0], time_idx[-1]))
#show the original time window of X in black, then
#show the subtracted reconstruction or the high mode
#reconstruction in red
if self.original_channels == 1:
plt.figure()
plt.rc('text', usetex=True)
plt.plot(time_idx, X[0, :], 'k', label=r'\left| X \right|')
if self.subtraction:
plt.plot(time_idx,
Xhatlow[0, :],
'r',
label=r'\left| \hat{X} \right|')
else:
plt.plot(time_idx,
Xhathigh[0, :],
'r',
label=r'\left| \hat{X} \right| left over modes')
plt.legend()
title = r'Reconstruction level: %d' % l
plt.title(title)
plt.show()
else:
self._day_plot(time_idx,
X,
title=r'\left| X \right|, level: %d' % l,
savefig=savefig)
if self.subtraction:
title_end = r'\left| \hat{X} \right|, level: %d' % l
self._day_plot(time_idx,
Xhatlow,
color='r',
title=title_end,
savefig=savefig)
elif self.excess_reconstruction:
title_end = r'\left| \hat{X} \right| left over modes, level: %d' % l
self._day_plot(time_idx,
Xhathigh,
color='r',
title=title_end,
savefig=savefig)
#show the reconstruction for the next levels only for
#subtraction constructed via: x - xhat
if self.subtraction:
if self.original_channels == 1:
plt.figure()
plt.rc('text', usetex=True)
plt.plot(time_idx,
X_next_l[0, :],
'b',
label=r'\left| X \right|')
title = r'\left| X - \hat{X} \right| level: %d' % l
plt.title(title)
plt.show()
if savefig:
plt.savefig('%s.pdf' % title)
plt.close()
else:
title = r'\left| X - \hat{X} \right| level: %d' % l
self._day_plot(time_idx,
X_next_l,
color='b',
title=title,
savefig=savefig)
#show heatmap so far
self.heatmap(title='Level %d' % l)
return tam, norm
def F1(w, t):
f1 = 3 * np.cos(w * t) + 2 * np.cos(3 * w * t) + np.cos(5 * w * t)
return f1
n = 51
T = 2 * np.pi
h = T / n
t = np.linspace(0, 50 * h, n)
w = 2 * np.pi / T
tam, val = fourier(F1(w, t))
tf = np.linspace(0, tam, tam)
plt.figure(1, figsize=(14, 5))
plt.subplot(1, 2, 1)
plt.plot(t, F1(w, t))
plt.scatter(t, F1(w, t), s=9)
plt.xlabel('t')
plt.ylabel('y(t)')
plt.subplot(1, 2, 2)
plt.stem(tf, abs(val), use_line_collection=True)
plt.xlabel('k')
plt.ylabel('|X|/N')
plt.savefig('1.png')
plt.close()
# variance of each feature
plt.figure()
plt.plot(featureList[i].Var)
plt.title('variance ' + featureName[i])
plt.xlabel('File number')
plt.savefig('plot/variance ' + featureName[i] + '.png', dpi=1000)
plt.close()
# plot distribution of some features
if i > 2:
distributionData = dataAll[:, i]
valueCounter = collections.Counter(distributionData)
plt.figure()
plt.stem(valueCounter.keys(), valueCounter.values())
plt.title(featureName[i] + ' distribution with zero')
plt.xlabel('Value')
plt.ylabel('Quantity')
plt.savefig('plot/' + featureName[i] +
' value distribution with zero.png',
dpi=1000)
plt.close()
plt.figure()
plt.stem(valueCounter.keys()[1:], valueCounter.values()[1:])
plt.xlabel('Value')
plt.ylabel('Quantity')
plt.title(featureName[i] + ' distribution without zero')
plt.savefig('plot/' + featureName[i] +
' distribution without zero.png',
dpi=1000)
def SplitWavdataByEnergy(sig, fs, initialsegmentfolder, file, Xsec):
# Splits wav data based on energy, i.e. pauses
# Input: sig: the wavfile data in an array
# fs: sampling frequency
# initialsegmentfolder: the folder location where all the wav segments are dumped
# file: the wavfile to be split into segments
# Xsec: the minimum duration of the segments that we'll split the wavfile
# Output: wavfile segments from 0 to N in initialsegmentfolder
#########################################################################
window = 512
hop = window / 2
energy = []
i = 0
energy_index = []
while i < (len(sig) - window):
chunk = sig[i:i + window][np.newaxis]
energy.append(chunk.dot(chunk.T)[0][0])
energy_index.append(i)
i = i + hop
energy = np.array(energy)
energy_thresh = 0.1 * np.mean(
energy) #mean because there might be some spurious peak in energy
indiceswithlowenergy = np.where(energy <= energy_thresh)
timeinstance_withlowenergy = indiceswithlowenergy[0] * hop * 1.0 / fs
### retain those silences which are greater than or equal to 0.2 seconds, and hence find valid silent segments
sil_dur_cap = 0.2
num_samp_sil_dur_cap = np.floor(sil_dur_cap * fs * 1.0 / hop)
lowenergyindices = indiceswithlowenergy[0]
validlowenergy_subarray = []
validlowenergyarray = []
print '~~~~num_samp_sil_dur_cap = ', num_samp_sil_dur_cap
print '\nlen(lowenergyindices): ', len(lowenergyindices)
for ind in range(len(lowenergyindices) - 1):
diff = lowenergyindices[ind + 1] - lowenergyindices[ind]
if diff > 1:
##to account for breathy regions## BUT THIS PIECE OF CODE SPLITS FROM CONSONANTS ## NOT desirable
# if diff>np.floor(0.2*fs*1.0/hop) and diff<np.floor(0.3*fs*1.0/hop): #0.2-0.3 seconds of breathy voice allowed
# for i in range(lowenergyindices[ind],lowenergyindices[ind+1],1):
# validlowenergy_subarray.append(i)
# continue
#################################
if validlowenergy_subarray:
validlowenergy_subarray.append(lowenergyindices[ind])
if len(validlowenergy_subarray) >= num_samp_sil_dur_cap:
validlowenergyarray = validlowenergyarray + validlowenergy_subarray
validlowenergy_subarray = []
continue
validlowenergy_subarray.append(lowenergyindices[ind])
if len(validlowenergy_subarray) >= num_samp_sil_dur_cap:
validlowenergyarray = validlowenergyarray + validlowenergy_subarray
validlowenergy_subarray = []
#########################
##Finding center of valid silent regions. These will be boundaries of phrases/segments/song lines
# print '\nlen(validlowenergyarray): ',len(validlowenergyarray)
boundary = []
for ind in range(len(validlowenergyarray) - 1):
diff = validlowenergyarray[ind + 1] - validlowenergyarray[ind]
if diff > 1:
if validlowenergy_subarray:
validlowenergy_subarray.append(validlowenergyarray[ind])
boundary.append(validlowenergy_subarray[0] +
((validlowenergy_subarray[-1] -
validlowenergy_subarray[0]) / 2))
validlowenergy_subarray = []
# print '\nI\'m before Continue. Current iter is: ',ind,'\n'
continue
validlowenergy_subarray.append(validlowenergyarray[ind])
if validlowenergy_subarray:
boundary.append(validlowenergy_subarray[0] + (
(validlowenergy_subarray[-1] - validlowenergy_subarray[0]) / 2))
print 'len(boundary): ', len(boundary)
WavSplitMinXsec(fs, initialsegmentfolder, file, sig, boundary, hop, Xsec)
##########################
if plot:
plt.figure("figure from SplitWavdataByEnergy 1")
plt.subplot(2, 1, 1)
plt.plot(range(len(sig)), sig)
plt.ylabel('amplitude')
plt.subplot(2, 1, 2)
plt.plot(energy_index, energy)
plt.stem(indiceswithlowenergy[0] * hop,
5 * np.ones(len(indiceswithlowenergy[0])), 'k')
plt.ylabel('energy')
plt.show()
plt.figure("figure from SplitWavdataByEnergy 2")
plt.title('amplitude vs. time')
plt.subplot(2, 1, 1)
plt.plot(np.array(range(len(sig))) * 1.0 / fs, sig)
plt.ylabel('amplitude')
plt.subplot(2, 1, 2)
plt.plot(np.array(energy_index) * 1.0 / fs, energy)
plt.stem(timeinstance_withlowenergy,
5 * np.ones(len(indiceswithlowenergy[0])), 'k')
plt.ylabel('energy')
plt.show()
if plot:
plt.figure("figure from SplitWavdataByEnergy 3")
plt.subplot(3, 1, 1)
plt.plot(range(len(sig)), sig)
plt.ylabel('amplitude')
plt.subplot(3, 1, 2)
plt.plot(energy_index, energy)
plt.stem(indiceswithlowenergy[0] * hop,
5 * np.ones(len(indiceswithlowenergy[0])), 'k')
plt.ylabel('energy')
plt.subplot(3, 1, 3)
plt.plot(energy_index, energy)
plt.stem(
np.array(validlowenergyarray) * hop,
10 * np.ones(len(validlowenergyarray)), 'k')
plt.ylabel('energy')
plt.show()
plt.figure("figure from SplitWavdataByEnergy 4")
plt.title('amplitude vs. time')
plt.subplot(2, 1, 1)
plt.plot(np.array(range(len(sig))) * 1.0 / fs, sig)
plt.ylabel('amplitude')
plt.subplot(2, 1, 2)
plt.plot(np.array(energy_index) * 1.0 / fs, energy)
plt.stem(
np.array(validlowenergyarray) * hop * 1.0 / fs,
5 * np.ones(len(validlowenergyarray)), 'k')
plt.ylabel('energy')
plt.show()
if plot:
plt.figure()
plt.title('amplitude vs. time')
plt.subplot(2, 1, 1)
plt.plot(np.array(range(len(sig))) * 1.0 / fs, sig)
plt.ylabel('amplitude')
plt.subplot(2, 1, 2)
plt.plot(np.array(energy_index) * 1.0 / fs, energy)
plt.stem(
np.array(validlowenergyarray) * hop * 1.0 / fs,
1 * np.ones(len(validlowenergyarray)), 'k')
plt.stem(
np.array(boundary) * hop * 1.0 / fs, 10 * np.ones(len(boundary)),
'r')
plt.ylabel('energy')
plt.show()
return