def plotMeans(numDice, numRolls, numBins, legend, color, style):
means = []
for i in range(numRolls // numDice):
vals = 0
for j in range(numDice):
vals += 5 * random.random()
means.append(vals / float(numDice))
numpy.hist(means,
numBins,
color=color,
label=legend,
weights=[1 / len(means)] * len(means),
hatch=style)
return getMeanAndStd(means)
def JacobianStats(inputImage, range1):
#input a jacobian image generated by ANTSJacobian and output histogram
#need to include values above maximum in range1
inputImage = imageDB.loadNifti(inputImage)
range1 = np.append(range1, np.max(inputImage))
counts, range1 = np.hist(inputImage, range1)
return counts
def segment(image, v_min=5, v_max=250):
# get rid of white and black pixels in histogram
bins, edges = np.hist(image.flatten(),
bins=v_max - v_min,
range=(v_min, v_max))
edges = edges.astype(np.int) # should be int
# identify grey peak
largest_bin = np.argmax(bins)
largest_edge = edges[largest_bin]
def PeakShape(): KK = np.floor(int(args.d)/int(args.w)) mu = int(args.d) r = 2. p = r/(mu+r) X = np.random.negative_binomial(r,p,1e6) X2 = np.ceil(X*np.random.rand(1,1e6)) I = np.nonzero(np.logical_and(X > 60, X < 170))[1] L1 = np.ceil((1-X2[0][I]/args.w) L2 = np.ceil((X[I]-X2[0][I])/args.w) H1 = np.histogram(L1,np.arange((-1000/int(args.w),0))) H2 = np.hist( L2,np.arange(0,(1000/int(args.w)))) H1 = H1/sum(H1) H2 = H2/sum(H2) dat1 = [H1 np.zeros(1000/int(args.w))] c1 = np.convolve(dat1,np.ones(KK)) c1 = c1[:-KK+1] dat2 = [np.zeros(1000/int(args.w)) H2] c2 = np.convolve(dat2,np.ones(KK)) c2 = c2[KK:] c = c1 + c2 c = smooth(c,5) c = c/max(c) return c def Extrap(): shape = shape[shape > 0] lshape = len(shape) shp = shape[shape > 0] shp = shp/max(shp) lshp = len(shp) shape2 = shape[shape > 0] shape2 = shape2/max(shape2) lshape2 = len(shape2) shape2 = np.interp(np.arange(1:int(args.w)*lshape2:int(args.w)),shape2,np.arange(1,jump*lshape2)) lshape2 = len(shape2) def InitData(): def PFR(): def FitShape(): shape = PeakShape() Extra(shape) InitData()
def get_map_range(self, channel):
# Check if channel is a string specifying the PFM channel to plot
if type(channel) is str:
if channel in self.channels.keys():
data = self.channels[channel]
else:
raise ValueError(
"Channel to plot needs to be one of {}".format(
self.channels.keys()))
# Check if it is the data array of the PFM channel to plot
elif isinstance(channel, np.ndarray):
if channel in self.channels.values():
data = channel
else:
raise ValueError(
"Channel data does not match loaded channels in class!")
data_hist = np.hist(data)
def normalized_hist(img_gray, num_bins):
assert len(img_gray.shape) == 2, 'image dimension mismatch'
assert img_gray.dtype == 'float', 'incorrect image type'
hists = np.zeros(num_bins)
bins = np.zeros(num_bins)
bin_len = 255.0 / num_bins
for i in range(num_bins):
bins[i] = bin_len * (i + 1)
bins = np.insert(bins, 0, 0)
for i in range(img_gray.shape[0]):
for j in range(img_gray.shape[1]):
k = min(int(img_gray[i, j] / bin_len), num_bins)
hists[k] += 1
hists, bins = hist(img_gray.reshape(img_gray.size), num_bins, (0, 255))
#Normalized
hists = hists * 1.0 / np.sum(hists)
print(hists, bins)
return hists, bins
def plot_vel_dir(alist, abs_dir):
velocity = get_params_as_array(alist,"Velocity")
pos = get_params_as_array(alist,"pos")
v = np.zeros(np.size(N))
i=0
for (vel,p) in zip(velocity,pos):
r = (p[0]**2 + p[1]**2 + p[2]**2)**0.5
v[i] = vel*p/r
i = i + 1
v = (v * yt.units.cm).convert_to_units('km').value #pf['km'] / pf['cm'] # now in km/s
hist, bins = np.hist(v, bins = 25, density =False)
centers = 0.5*(bins[:-1] + bins[1:])
plt.plot(centers, hist)
plt.xlabel('vel projected towards cluster center')
plt.close()
def boutlength_distribution(df, _bins=np.arange(0,10,100)):
dist = np.hist(df['bout_duration'], bins=_bins)[0]
count = sum(dist)
return dist/float(count)
def makeHist(data, title, xlabel, ylabel, bins=20):
numpy.hist(data, bins=bins)
numpy.title(title)
numpy.xlabel(xlabel)
numpy.ylabel(ylabel)
bin_nums = 500
initial_guess = [51, 1, 11, 25, 1, 13, 15, 1, 17]
# hint [height1, width1, peak position1,height2, width2, peak position2, height3, width3, peak position3]
############################################
#read the imput file
with open(imp_file, "r") as f:
dataset = []
for line in f:
num = float(line)
dataset.append(num)
#make histogram with 500 bins (a tuple is created)
h = hist(dataset,
bins=bin_nums,
range=None,
normed=False,
weights=None,
density=None)
#divide a tuple into separate strings
x = h[1]
xdata = x[:-1]
ydata = h[0]
#plot histogram
plt.plot(xdata, ydata, 'b-', label='data')
#define fitting function: 3 gaussians in this case
def func(x, a1, b1, c1, a2, b2, c2, a3, b3, c3):
return a1 * np.exp(-b1 * (x - c1)**2) + a2 * np.exp(
Function that returns log of truncated at [a, b] normal distribution with
mean ``mean`` and variance ``sigma**2``.
"""
x_ = np.where((a < x) & (x < b), x, 1)
k = math.log(norm.cdf((b - mean) / sigma) -
norm.cdf((a - mean) / sigma))
result1 = -math.log(sigma) - 0.5 * math.log(2. * math.pi) -\
0.5 * ((x_ - mean) / sigma) ** 2. - k
result = np.where((a < x) & (x < b), result1, float("-inf"))
return result
if __name__ == '__main__':
dgamma = Dlognorm_shifted(2.)
np.hist(dgamma(mean=2, sigma=0.5, size=10000), bins=50, normed=True)
np.hist(get_theta_sample(dgamma, mean=2, sigma=0.5, size=10000) * 180. /
math.pi, bins=50, normed=True)
# Load data
data_lists = list()
files = ['source.txt', 'beta.txt', 'sigma_beta.txt']
for file_ in files:
with open(file_) as f:
data_lists.append(f.read().splitlines())
sources_list = data_lists[0]
betas_list = data_lists[1]
beta_sigmas_list = data_lists[2]
# Convert to float numbers
for i, value in enumerate(betas_list):
def generate_figures():
header_data = np.loadtxt(dev_const.PUB_CONDEL_PREDICTION_RESULT,
dtype='S20'
)[:, :5]
scores_data = np.loadtxt(dev_const.PUB_CONDEL_PREDICTION_RESULT,
dtype='S20'
)[:, 5:13].astype(np.float)
min_value = np.amin(scores_data)
max_value = np.amax(scores_data)
predictor_names = ('CombiVEP',
'Phylop',
'SIFT',
'PP2',
'LRT',
'MT',
'GERP',
'Condel',
)
predictor_colors = ('k',
'm',
'c',
'g',
'b',
'coral',
'darkred',
'r',
)
#produce roc data from CombiVEP, Phylop, SIFT, PP2, LRT, MT, GERP, Condel
fp_rates, tp_rates = calculate_roc(scores_data[header_data[:, 4] == '1'],
scores_data[header_data[:, 4] == '0'],
np.linspace(min_value, max_value, 5001))
fig = plt.figure()
ax = fig.add_subplot(111)
for i in xrange(len(predictor_names)):
ax.plot(fp_rates[:, i],
tp_rates[:, i],
predictor_colors[i],
label=predictor_names[i])
ax.set_ylabel('true positive rate')
ax.set_xlabel('false positive rate')
ax.legend(bbox_to_anchor=(0.9999, 0.0001), loc=4)
fig.savefig(dev_const.PUB_ROC_FIG, bbox_inches='tight', pad_inches=0.05)
#produce auc data from roc data
fig = plt.figure()
aucs = []
ind = []
ax = fig.add_subplot(111)
for i in xrange(len(predictor_names)):
aucs.append(auc(fp_rates[:, i], tp_rates[:, i]))
ind.append(0.5*(i+1)-0.4)
ax.bar(ind, aucs, 0.3, color=predictor_colors)
for i in xrange(len(aucs)):
ax.text(ind[i], aucs[i] + 0.01, "%0.3f" % aucs[i])
ax.set_ylim([0.7, 0.9])
ax.set_xticks(np.array(ind) + 0.15)
ax.set_xticklabels(predictor_names, rotation=30)
fig.savefig(dev_const.PUB_AUC_FIG, bbox_inches='tight', pad_inches=0.05)
#plot scores distribution
fig = plt.figure()
ax = fig.add_subplot(211)
hist_range = (-0.005, 1.005)
patho_hist, bins = hist(scores_data[header_data[:, 4] == '1'][:, 0],
bins=100,
range=hist_range)
neutr_hist, bins = hist(scores_data[header_data[:, 4] == '0'][:, 0],
bins=100,
range=hist_range)
center = (bins[:-1]+bins[1:]) / 2
ax.plot(center, patho_hist, 'r--', label='pathogenic variants')
ax.plot(center, neutr_hist, 'b--', label='neutral variants')
ax.set_title('CombiVEP score distributuion')
ax.set_ylabel('samples')
ax.set_xlabel('score')
ax.legend(bbox_to_anchor=(0.999, 0.999), loc=1)
ax = fig.add_subplot(212)
patho_hist, bins = hist(scores_data[header_data[:, 4] == '1'][:, 7],
bins=100,
range=hist_range)
neutr_hist, bins = hist(scores_data[header_data[:, 4] == '0'][:, 7],
bins=100,
range=hist_range)
center = (bins[:-1]+bins[1:])/2
ax.plot(center, patho_hist, 'r--', label='pathogenic variants')
ax.plot(center, neutr_hist, 'b--', label='neutral variants')
ax.set_title('Condel score distributuion')
ax.set_ylabel('samples')
ax.set_xlabel('score')
ax.legend(bbox_to_anchor=(0.999, 0.999), loc=1)
fig.tight_layout()
fig.savefig(dev_const.PUB_SCORES_DISTR_FIG,
bbox_inches='tight',
pad_inches=0.05)
return (dev_const.PUB_ROC_FIG,
dev_const.PUB_AUC_FIG,
dev_const.PUB_SCORES_DISTR_FIG,
)
return gray
## gray-value histograms (Question 2.a)
img_color = np.array(Image.open('./model/obj100__0.png'))
img_gray = rgb2gray(img_color.astype('double'))
plt.figure(1)
plt.subplot(1, 3, 1)
plt.imshow(img_color)
plt.subplot(1, 3, 2)
num_bins_gray = 40
hist_gray1, bin_gray1 = hist(img_gray.reshape(img_gray.size), num_bins_gray,
(0, 255))
plt.bar((bin_gray1[0:-1] + bin_gray1[1:]) / 2, hist_gray1)
plt.subplot(1, 3, 3)
hist_gray2, bin_gray2 = histogram_module.normalized_hist(
img_gray, num_bins_gray)
plt.bar((bin_gray2[0:-1] + bin_gray2[1:]) / 2, hist_gray2)
plt.tight_layout()
## more histograms (Question 2.b)
#Compose and test RGB histograms (histogram_module.rgb_hist)
plt.figure(2)
plt.subplot(1, 2, 1)
plt.imshow(img_color)
def _hist_bins(self, lens) -> np.ndarray:
return np.r_[0, np.hist(lens, self.bins_number)[1]]
dy = np.diff(get(ax[i,1], 'ylim'))*inset
set(ax[i,:], 'ylim', [ylimmin[i,1)-dy, ylimmax[i,1]+dy])
dx = np.zeros([1,cols])
for j in range(cols):
set(ax[1,j], 'xlim', [xlimmin[1,j], xlimmax[1,j]])
dx(j) = np.diff(get(ax[1,j], 'xlim'))*inset
set(ax[:,j],'xlim',[xlimmin[1,j]-dx[j] xlimmax[1,j]+dx[j]])
set(ax[1:rows-1, :], 'xticklabel', '')
set(ax[:, 2:cols], 'yticklabel', '')
set(BigAx,'XTick', get(ax[rows,1], 'xtick'), 'YTick', get(ax[rows,1], 'ytick'), 'userdata', ax, 'tag', 'PlotMatrixBigAx')
if dohist: # Put a histogram on the diagonal for plotmatrix(y) case
for i in range(rows:-1:1):
histax = axes('Position', get(ax[i,i],'Position'), 'HandleVisibility', BigAxHV, 'parent', BigAxParent)
nn,xx = np.hist(np.reshape(y[:,i,:],[m,k]))
patches[i,:] = plt.bar(histax, xx, nn, 'hist')
if putlabels:
xt = 0.5*(np.max(y[:,i,:])+np.min(y[:,i,:]))
yt = 0.9*np.max(nn)
txt = varnames[i]
plt.text(xt, yt, txt)
set(histax, 'xtick', [], 'ytick', [], 'xgrid', 'off', 'ygrid', 'off')
set(histax, 'xlim', [xlimmin[1,i]-dx[i], xlimmax[1,i]+dx[i]])
pax[i] = histax # ax handles for histograms
patches = patches
# A bug seems to occur when plotmatrix is ran to produce a plot inside a GUI
# whereby the default fig menu items and icons appear. Commenting out the code below fixed the issue.
# Tim Peterson - April 2016
# Make BigAx the CurrentAxes
filepath = path+file0
file1 = 'rhow_'+label+'_' + tlabel
fileout = path + file1
#
print 'opening: ', filepath
#
#
data = vtktools.vtu(filepath)
print 'fields: ', data.GetFieldNames()
print 'extract V, R'
V = data.GetVectorField('Velocity_CG')
R = data.GetScalarField('Density_CG')
rho = []
w = []
for d in range(len(depths)):
rho.append(np.hist(R[coords[:,2]==depths[d]]))
w.append(np.mean(V[coords[:,2]==depths[d],2]))
rho = np.asarray(rho)
w = np.asarray(w)
#del data
#
print 'max: ', (w*rho).max(), 'min: ', (w*rho).min()
#
# RHO*W
fig = plt.figure(figsize=(2,5))
ax = fig.add_subplot(111)
rhow_z = rho*w
plt.plot([0, 0], [min(depths), max(depths)], color='k', linestyle='--', linewidth=1)
plt.plot(rhow_z,depths,color='0.75')
plt.xlabel('$<w*rho>$')
clf()
R.get_group( 63).intens5.hist(bins=linspace(0,200,1000), normed=False, log=1, histtype='step')
R.get_group( 63).intens5.hist(bins=linspace(0,100,1000), normed=False, log=1, histtype='step')
df_r.intens5.max()
df_r.intens5.min()
df_r.intens5.median()
R.get_group( 63).intens5.hist(bins=linspace(0,80,100), normed=False, log=1, histtype='step')
R.get_group( 120).intens5.hist(bins=linspace(0,80,100), normed=False, log=1, histtype='step')
runs
R.get_group( 128).intens5.hist(bins=linspace(0,80,100), normed=False, log=1, histtype='step')
R.get_group( 66).intens5.hist(bins=linspace(0,80,100), normed=False, log=1, histtype='step')
#R.get_group( 66).intens5.hist(bins=linspace(0,80,100), normed=False, log=1, histtype='step')
df_r / df_r.partiality
df_r.intens5 / df_r.partiality
np.sqrt( df_r.intens5 / df_r.partiality )
np.hist(np.sqrt( df_r.intens5 / df_r.partiality ) )
(np.sqrt( df_r.intens5 / df_r.partiality ) ).hist(bins=100)
df
df.BnotA
np.all(df.BnotA)
np.all(df.AandB)
np.aany(df.AandB)
np.any(df.AandB)
#df.inten5/ df.partiality
df = df.query("HS_ratio < 0.9")
df
df = df.query("HS_ratio < 0.9")
df
df.intens5 / df.partiality
df.intens5 / df.partialitB
df.intens5 / df.partialityB
def cdf_benchmarking_plot(timeseries, kernel_density=False):
""" Plot probability distribution of model outputs """
combined_df = pd.DataFrame()
for ts in timeseries:
df1 = ts.tp.to_dataframe(name=ts.plot_legend)
df2 = df1.dropna().reset_index()
df3 = df2.drop(["time", "lon", "lat"], axis=1)
combined_df[ts.plot_legend] = df3[ts.plot_legend]
time_ds = timeseries[0].time.to_dataframe(name='time')
months_float = np.ceil((time_ds["time"] - np.floor(time_ds["time"])) * 12)
df = combined_df.copy()
df["time"] = months_float.values
df["time"] = df["time"].astype(int)
grouped_dfs = []
for m in range(1, 13):
month_df = df[df['time'] == m]
month_df = month_df.drop(['time'], axis=1)
grouped_dfs.append(month_df)
# Plot
_fig, axs = plt.subplots(3, 4, sharex=True, sharey=True)
for i in range(12):
x = (i) % 3
y = int(i / 3)
for b in list(combined_df):
data = grouped_dfs[i][b].values
X_plot = np.linspace(0, data.max(), 1000)[:, np.newaxis]
# data_sorted = np.sort(data)
# p = 1. * np.arange(len(data)) / (len(data) - 1)
if kernel_density is True:
n, _bins, _patches = np.hist((grouped_dfs[i])[b],
cumulative=True,
density=True,
label=b)
X = n.reshape(-1, 1)
kde = KernelDensity(kernel='gaussian', bandwidth=0.1).fit(X)
log_dens = kde.score_samples(X_plot)
axs[x, y].plot(X_plot[:, 0], np.exp(log_dens), label=b)
else:
axs[x, y].hist((grouped_dfs[i])[b],
cumulative=True,
density=True,
label=b)
axs[x, y].set_title(month_dict[i])
axs[x, y].set_title(month_dict[i])
axs[x, y].xaxis.set_tick_params(which="both", labelbottom=True)
axs[x, y].yaxis.set_tick_params(which="both", labelbottom=True)
axs[x, y].set_xlabel("Precipitation (mm/day)")
axs[x, y].set_ylabel("CDF")
axs[x, y].axvline(
np.percentile((grouped_dfs[i])['ERA5'], 95),
color="k",
linestyle="dashed",
linewidth=1,
label="ERA5 95th percentile",
)
plt.legend(loc="upper right")
plt.show()
random_state=0,
stratify=None)
qtrans = QuantileTransformer2(numerical_features='auto', drop=True, dtype=np.float32, n_quantiles=1000,
output_distribution='uniform', ignore_implicit_zeros=False, subsample=100000,
random_state=42, copy=True)
X_trans = qtrans.fit_transform(X_train)
print(X_trans.info())
# show X_train data
X_trans.hist(bins=50, figsize=(10, 10))
#plt.show()
X_train.hist(bins=50, figsize=(10, 10))
#plt.show()
X_trans.to_excel("output.xlsx")
print(X_train.skew(numeric_only=True),'\n')
print(X_trans.skew(numeric_only=True))
input=X_trans['MSSubClass_qnt'].to_numpy().flatten()
a=1
b=8
output=beta.pdf(input, a, b, loc=0, scale=1)
print(f'output type: {type(output)}')
print('skew=',skew(output,axis=0))
print(np.hist(output))
