I've got the following simple script that plots a graph:
import matplotlib.pyplot as plt
import numpy as np
T = np.array([6, 7, 8, 9, 10, 11, 12])
power = np.array([1.53E+03, 5.92E+02, 2.04E+02, 7.24E+01, 2.72E+01, 1.10E+01, 4.70E+00])
plt.plot(T,power)
plt.show()
As it is now, the line goes straight from point to point which looks ok, but could be better in my opinion. What I want is to smooth the line between the points. In Gnuplot I would have plotted with smooth cplines.
Is there an easy way to do this in PyPlot? I've found some tutorials, but they all seem rather complex.
You could use scipy.interpolate.spline to smooth out your data yourself:
from scipy.interpolate import spline
# 300 represents number of points to make between T.min and T.max
xnew = np.linspace(T.min(), T.max(), 300)
power_smooth = spline(T, power, xnew)
plt.plot(xnew,power_smooth)
plt.show()
spline is deprecated in scipy 0.19.0, use BSpline class instead.
Switching from spline to BSpline isn't a straightforward copy/paste and requires a little tweaking:
from scipy.interpolate import make_interp_spline, BSpline
# 300 represents number of points to make between T.min and T.max
xnew = np.linspace(T.min(), T.max(), 300)
spl = make_interp_spline(T, power, k=3) # type: BSpline
power_smooth = spl(xnew)
plt.plot(xnew, power_smooth)
plt.show()
Before:
After:
For this example spline works well, but if the function is not smooth inherently and you want to have smoothed version you can also try:
from scipy.ndimage.filters import gaussian_filter1d
ysmoothed = gaussian_filter1d(y, sigma=2)
plt.plot(x, ysmoothed)
plt.show()
if you increase sigma you can get a more smoothed function.
Proceed with caution with this one. It modifies the original values and may not be what you want.
See the scipy.interpolate documentation for some examples.
The following example demonstrates its use, for linear and cubic spline interpolation:
import matplotlib.pyplot as plt
import numpy as np
from scipy.interpolate import interp1d
# Define x, y, and xnew to resample at.
x = np.linspace(0, 10, num=11, endpoint=True)
y = np.cos(-x**2/9.0)
xnew = np.linspace(0, 10, num=41, endpoint=True)
# Define interpolators.
f_linear = interp1d(x, y)
f_cubic = interp1d(x, y, kind='cubic')
# Plot.
plt.plot(x, y, 'o', label='data')
plt.plot(xnew, f_linear(xnew), '-', label='linear')
plt.plot(xnew, f_cubic(xnew), '--', label='cubic')
plt.legend(loc='best')
plt.show()
Slightly modified for increased readability.
One of the easiest implementations I found was to use that Exponential Moving Average the Tensorboard uses:
def smooth(scalars: List[float], weight: float) -> List[float]: # Weight between 0 and 1
last = scalars[0] # First value in the plot (first timestep)
smoothed = list()
for point in scalars:
smoothed_val = last * weight + (1 - weight) * point # Calculate smoothed value
smoothed.append(smoothed_val) # Save it
last = smoothed_val # Anchor the last smoothed value
return smoothed
ax.plot(x_labels, smooth(train_data, .9), x_labels, train_data)
I presume you mean curve-fitting and not anti-aliasing from the context of your question. PyPlot doesn't have any built-in support for this, but you can easily implement some basic curve-fitting yourself, like the code seen here, or if you're using GuiQwt it has a curve fitting module. (You could probably also steal the code from SciPy to do this as well).
Here is a simple solution for dates:
from scipy.interpolate import make_interp_spline
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates as dates
from datetime import datetime
data = {
datetime(2016, 9, 26, 0, 0): 26060, datetime(2016, 9, 27, 0, 0): 23243,
datetime(2016, 9, 28, 0, 0): 22534, datetime(2016, 9, 29, 0, 0): 22841,
datetime(2016, 9, 30, 0, 0): 22441, datetime(2016, 10, 1, 0, 0): 23248
}
#create data
date_np = np.array(list(data.keys()))
value_np = np.array(list(data.values()))
date_num = dates.date2num(date_np)
# smooth
date_num_smooth = np.linspace(date_num.min(), date_num.max(), 100)
spl = make_interp_spline(date_num, value_np, k=3)
value_np_smooth = spl(date_num_smooth)
# print
plt.plot(date_np, value_np)
plt.plot(dates.num2date(date_num_smooth), value_np_smooth)
plt.show()
It's worth your time looking at seaborn for plotting smoothed lines.
The seaborn lmplot function will plot data and regression model fits.
The following illustrates both polynomial and lowess fits:
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
T = np.array([6, 7, 8, 9, 10, 11, 12])
power = np.array([1.53E+03, 5.92E+02, 2.04E+02, 7.24E+01, 2.72E+01, 1.10E+01, 4.70E+00])
df = pd.DataFrame(data = {'T': T, 'power': power})
sns.lmplot(x='T', y='power', data=df, ci=None, order=4, truncate=False)
sns.lmplot(x='T', y='power', data=df, ci=None, lowess=True, truncate=False)
The order = 4 polynomial fit is overfitting this toy dataset. I don't show it here but order = 2 and order = 3 gave worse results.
The lowess = True fit is underfitting this tiny dataset but may give better results on larger datasets.
Check the seaborn regression tutorial for more examples.
Another way to go, which slightly modifies the function depending on the parameters you use:
from statsmodels.nonparametric.smoothers_lowess import lowess
def smoothing(x, y):
lowess_frac = 0.15 # size of data (%) for estimation =~ smoothing window
lowess_it = 0
x_smooth = x
y_smooth = lowess(y, x, is_sorted=False, frac=lowess_frac, it=lowess_it, return_sorted=False)
return x_smooth, y_smooth
That was better suited than other answers for my specific application case.
Related
I've got the following simple script that plots a graph:
import matplotlib.pyplot as plt
import numpy as np
T = np.array([6, 7, 8, 9, 10, 11, 12])
power = np.array([1.53E+03, 5.92E+02, 2.04E+02, 7.24E+01, 2.72E+01, 1.10E+01, 4.70E+00])
plt.plot(T,power)
plt.show()
As it is now, the line goes straight from point to point which looks ok, but could be better in my opinion. What I want is to smooth the line between the points. In Gnuplot I would have plotted with smooth cplines.
Is there an easy way to do this in PyPlot? I've found some tutorials, but they all seem rather complex.
You could use scipy.interpolate.spline to smooth out your data yourself:
from scipy.interpolate import spline
# 300 represents number of points to make between T.min and T.max
xnew = np.linspace(T.min(), T.max(), 300)
power_smooth = spline(T, power, xnew)
plt.plot(xnew,power_smooth)
plt.show()
spline is deprecated in scipy 0.19.0, use BSpline class instead.
Switching from spline to BSpline isn't a straightforward copy/paste and requires a little tweaking:
from scipy.interpolate import make_interp_spline, BSpline
# 300 represents number of points to make between T.min and T.max
xnew = np.linspace(T.min(), T.max(), 300)
spl = make_interp_spline(T, power, k=3) # type: BSpline
power_smooth = spl(xnew)
plt.plot(xnew, power_smooth)
plt.show()
Before:
After:
For this example spline works well, but if the function is not smooth inherently and you want to have smoothed version you can also try:
from scipy.ndimage.filters import gaussian_filter1d
ysmoothed = gaussian_filter1d(y, sigma=2)
plt.plot(x, ysmoothed)
plt.show()
if you increase sigma you can get a more smoothed function.
Proceed with caution with this one. It modifies the original values and may not be what you want.
See the scipy.interpolate documentation for some examples.
The following example demonstrates its use, for linear and cubic spline interpolation:
import matplotlib.pyplot as plt
import numpy as np
from scipy.interpolate import interp1d
# Define x, y, and xnew to resample at.
x = np.linspace(0, 10, num=11, endpoint=True)
y = np.cos(-x**2/9.0)
xnew = np.linspace(0, 10, num=41, endpoint=True)
# Define interpolators.
f_linear = interp1d(x, y)
f_cubic = interp1d(x, y, kind='cubic')
# Plot.
plt.plot(x, y, 'o', label='data')
plt.plot(xnew, f_linear(xnew), '-', label='linear')
plt.plot(xnew, f_cubic(xnew), '--', label='cubic')
plt.legend(loc='best')
plt.show()
Slightly modified for increased readability.
One of the easiest implementations I found was to use that Exponential Moving Average the Tensorboard uses:
def smooth(scalars: List[float], weight: float) -> List[float]: # Weight between 0 and 1
last = scalars[0] # First value in the plot (first timestep)
smoothed = list()
for point in scalars:
smoothed_val = last * weight + (1 - weight) * point # Calculate smoothed value
smoothed.append(smoothed_val) # Save it
last = smoothed_val # Anchor the last smoothed value
return smoothed
ax.plot(x_labels, smooth(train_data, .9), x_labels, train_data)
I presume you mean curve-fitting and not anti-aliasing from the context of your question. PyPlot doesn't have any built-in support for this, but you can easily implement some basic curve-fitting yourself, like the code seen here, or if you're using GuiQwt it has a curve fitting module. (You could probably also steal the code from SciPy to do this as well).
Here is a simple solution for dates:
from scipy.interpolate import make_interp_spline
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates as dates
from datetime import datetime
data = {
datetime(2016, 9, 26, 0, 0): 26060, datetime(2016, 9, 27, 0, 0): 23243,
datetime(2016, 9, 28, 0, 0): 22534, datetime(2016, 9, 29, 0, 0): 22841,
datetime(2016, 9, 30, 0, 0): 22441, datetime(2016, 10, 1, 0, 0): 23248
}
#create data
date_np = np.array(list(data.keys()))
value_np = np.array(list(data.values()))
date_num = dates.date2num(date_np)
# smooth
date_num_smooth = np.linspace(date_num.min(), date_num.max(), 100)
spl = make_interp_spline(date_num, value_np, k=3)
value_np_smooth = spl(date_num_smooth)
# print
plt.plot(date_np, value_np)
plt.plot(dates.num2date(date_num_smooth), value_np_smooth)
plt.show()
It's worth your time looking at seaborn for plotting smoothed lines.
The seaborn lmplot function will plot data and regression model fits.
The following illustrates both polynomial and lowess fits:
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
T = np.array([6, 7, 8, 9, 10, 11, 12])
power = np.array([1.53E+03, 5.92E+02, 2.04E+02, 7.24E+01, 2.72E+01, 1.10E+01, 4.70E+00])
df = pd.DataFrame(data = {'T': T, 'power': power})
sns.lmplot(x='T', y='power', data=df, ci=None, order=4, truncate=False)
sns.lmplot(x='T', y='power', data=df, ci=None, lowess=True, truncate=False)
The order = 4 polynomial fit is overfitting this toy dataset. I don't show it here but order = 2 and order = 3 gave worse results.
The lowess = True fit is underfitting this tiny dataset but may give better results on larger datasets.
Check the seaborn regression tutorial for more examples.
Another way to go, which slightly modifies the function depending on the parameters you use:
from statsmodels.nonparametric.smoothers_lowess import lowess
def smoothing(x, y):
lowess_frac = 0.15 # size of data (%) for estimation =~ smoothing window
lowess_it = 0
x_smooth = x
y_smooth = lowess(y, x, is_sorted=False, frac=lowess_frac, it=lowess_it, return_sorted=False)
return x_smooth, y_smooth
That was better suited than other answers for my specific application case.
I have a Unix time series (x) with an associated signal value (y) which is generated every minute, dropping the first value and appending a new one. I am trying to smooth the resulting curve without loosing time accuracy with a specific emphasis on the final value of the smoothed curve which will be written to a database. I would like to be able to adjust the smoothing to a considerable degree.
I have studied (as mathematical layman, more or less) all options I could find and I could master. I came across Savitzki Golay which looked perfect until I realized it works well on past data but fails to produce a reliable final value if no future data is available for smoothing. I have tried many other methods which produced results but could not be adjusted like Savgol.
import pandas as pd
from bokeh.plotting import figure, show, output_file
from bokeh.layouts import column
from math import pi
from scipy.signal import savgol_filter
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from scipy.interpolate import splrep, splev
from scipy.ndimage import gaussian_filter1d
from scipy.signal import lfilter
from scipy.interpolate import UnivariateSpline
import matplotlib.pyplot as plt
df_sim = pd.read_csv("/home/20190905_Signal_Smooth_Test.csv")
#sklearn Polynomial*****************************************
poly = PolynomialFeatures(degree=4)
X = df_sim.iloc[:, 0:1].values
print(X)
y = df_sim.iloc[:, 1].values
print(y)
X_poly = poly.fit_transform(X)
poly.fit(X_poly, y)
lin2 = LinearRegression()
lin2.fit(X_poly, y)
# Visualising the Polynomial Regression results
plt.scatter(X, y, color='blue')
plt.plot(X, lin2.predict(poly.fit_transform(X)), color='red')
plt.title('Polynomial Regression')
plt.xlabel('Time')
plt.ylabel('Signal')
plt.show()
#scipy interpolate********************************************
bspl = splrep(df_sim['timestamp'], df_sim['signal'], s=5)
bspl_y = splev(df_sim['timestamp'], bspl)
df_sim['signal_spline'] = bspl_y
#scipy gaussian filter****************************************
smooth = gaussian_filter1d(df_sim['signal'], 3)
df_sim['signal_gauss'] = smooth
#scipy lfilter************************************************
n = 5 # the larger n is, the smoother curve will be
b = [1.0 / n] * n
a = 1
histo_filter = lfilter(b, a, df_sim['signal'])
df_sim['signal_lfilter'] = histo_filter
print(df_sim)
#scipy UnivariateSpline**************************************
s = UnivariateSpline(df_sim['timestamp'], df_sim['signal'], s=5)
xs = df_sim['timestamp']
ys = s(xs)
df_sim['signal_univariante'] = ys
#scipy savgol filter****************************************
sg = savgol_filter(df_sim['signal'], 11, 3)
df_sim['signal_savgol'] = sg
df_sim['date'] = pd.to_datetime(df_sim['timestamp'], unit='s')
#plotting it all********************************************
print(df_sim)
w = 60000
TOOLS = "pan,wheel_zoom,box_zoom,reset,save"
p = figure(x_axis_type="datetime", tools=TOOLS, plot_width=1000, plot_height=250,
title=f"Various Signals y vs Timestamp x")
p.xaxis.major_label_orientation = pi / 4
p.grid.grid_line_alpha = 0.9
p.line(x=df_sim['date'], y=df_sim['signal'], color='green')
p.line(x=df_sim['date'], y=df_sim['signal_spline'], color='blue')
p.line(x=df_sim['date'], y=df_sim['signal_gauss'], color='red')
p.line(x=df_sim['date'], y=df_sim['signal_lfilter'], color='magenta')
p.line(x=df_sim['date'], y=df_sim['signal_univariante'], color='yellow')
p1 = figure(x_axis_type="datetime", tools=TOOLS, plot_width=1000, plot_height=250,
title=f"Savgol vs Signal")
p1.xaxis.major_label_orientation = pi / 4
p1.grid.grid_line_alpha = 0.9
p1.line(x=df_sim['date'], y=df_sim['signal'], color='green')
p1.line(x=df_sim['date'], y=df_sim['signal_savgol'], color='blue')
output_file("signal.html", title="Signal Test")
show(column(p, p1)) # open a browser
I expect a result that is similar to Savitzky Golay but with valid final smoothed values for the data series. None of the other methods present the same flexibility to adjust the grade of smoothing. Most other methods shift the curve to the right. I can provide to csv file for testing.
This really depends on why you are smoothing the data. Every smoothing method will have side effects, such as letting some 'noise' through more than other. Research 'phase response of filtering'.
A common technique to avoid the problem of missing data at the end of a symmetric filter is to just forecast your data a few points ahead and use that. For example, if you are using a 5-term moving average filter you will be missing 2 data points when you go to calculate your end value.
To forecast these two points, you could use the auto_arima() function from the pmdarima module, or look at the fbprophet module (which I find quite good for this kind of situation).
I have 3 data points on the x axis and 3 on the y axis:
x = [1,3,5]
y=[0,5,0]
I would like a curved line that starts at (1,0), goes to the highest point at (3,5) and then finishes at (5,0)
I think I need to use interpolation, but unsure how. If I use spline from scipy like this:
import bokeh.plotting as bk
from scipy.interpolate import spline
p = bk.figure()
xvals=np.linspace(1, 5, 10)
y_smooth = spline(x,y,xvals)
p.line(xvals, y_smooth)
bk.show(p)
I get the highest point before (3,5) and it looks unbalanced:
The issue is due to that spline with no extra argument is of order 3. That means that you do not have points/equations enough to get a spline curve (which manifests itself as a warning of an ill-conditioned matrix). You need to apply a spline of lower order, such as a cubic spline, which is of order 2:
import bokeh.plotting as bk
from scipy.interpolate import spline
p = bk.figure()
xvals=np.linspace(1, 5, 10)
y_smooth = spline(x,y,xvals, order=2) # This fixes your immediate problem
p.line(xvals, y_smooth)
bk.show(p)
In addition, spline is deprecated in SciPy, so you should preferably not use it, even if it is possible. A better solution is to use the CubicSpline class:
import bokeh.plotting as bk
from scipy.interpolate import CubicSpline
p = bk.figure()
xvals=np.linspace(1, 5, 10)
spl = CubicSpline(x, y) # First generate spline function
y_smooth = spl(xvals) # then evalute for your interpolated points
p.line(xvals, y_smooth)
bk.show(p)
Just to show the difference (using pyplot):
As can be seen, the CubicSpline is identical to the spline of order=2
use pchip_interpolate():
import numpy as np
from scipy import interpolate
x = [1,3,5]
y=[0,5,0]
x2 = np.linspace(x[0], x[-1], 100)
y2 = interpolate.pchip_interpolate(x, y, x2)
pl.plot(x2, y2)
pl.plot(x, y, "o")
the result:
You can use quadratic interpolation. This is possible by making use of scipy.interpolate.interp1d.
import matplotlib.pyplot as plt
from scipy.interpolate import interp1d
import numpy as np
x = [1, 3, 5]
y = [0, 5, 0]
f = interp1d(x, y, kind='quadratic')
x_interpol = np.linspace(1, 5, 1000)
y_interpol = f(x_interpol)
plt.plot(x_interpol, y_interpol)
plt.show()
Check the documentation for more details.
I am trying to interpolate spectrogram obtained from matplotlib using scipy's inetrp2d function, but somehow fail to get the same spectrogram. The data is available here
The actual spectrogram is:
And interpolated spectrogram is:
The code looks okay, but even then something is wrong. The code used is:
from __future__ import division
from matplotlib import ticker as mtick
from matplotlib.backends.backend_pdf import PdfPages
import matplotlib.pyplot as plt
import numpy as np
from bisect import bisect
from scipy import interpolate
from matplotlib.ticker import MaxNLocator
data = np.genfromtxt('spectrogram.dat', skiprows = 2, delimiter = ',')
pressure = data[:, 1] * 0.065
time = data[:, 0]
cax = plt.specgram(pressure * 100000, NFFT = 256, Fs = 50000, noverlap=4, cmap=plt.cm.gist_heat, zorder = 1)
f = interpolate.interp2d(cax[2], cax[1], cax[0], kind='cubic')
xnew = np.linspace(cax[2][0], cax[2][-1], 100)
ynew = np.linspace(cax[1][0], cax[1][-1], 100)
znew = 10 * np.log10(f(xnew, ynew))
fig = plt.figure(figsize=(6, 3.2))
ax = fig.add_subplot(111)
ax.set_title('colorMap')
plt.pcolormesh(xnew, ynew, znew, cmap=plt.cm.gist_heat)
# plt.colorbar()
plt.title('Interpolated spectrogram')
plt.colorbar(orientation='vertical')
plt.savefig('interp_spectrogram.pdf')
How to interpolate a spectrogram correctly with Python?
The key to your solution is in this warning, which you may or may not have seen:
RuntimeWarning: invalid value encountered in log10
znew = 10 * np.log10(f(xnew, ynew))
If your data is actually a power whose log you'd like to view explicitly as decibel power, take the log first, before fitting to the spline:
spectrum, freqs, t, im = cax
dB = 10*np.log10(spectrum)
#f = interpolate.interp2d(t, freqs, dB, kind='cubic') # docs for this recommend next line
f = interpolate.RectBivariateSpline(t, freqs, dB.T) # but this uses xy not ij, hence the .T
xnew = np.linspace(t[0], t[-1], 10*len(t))
ynew = np.linspace(freqs[0], freqs[-1], 10*len(freqs)) # was it wider spaced than freqs on purpose?
znew = f(xnew, ynew).T
Then plotting as you have:
Previous answer:
If you just want to plot on logscale, use matplotlib.colors.LogNorm
znew = f(xnew, ynew) # Don't take the log here
plt.figure(figsize=(6, 3.2))
plt.pcolormesh(xnew, ynew, znew, cmap=plt.cm.gist_heat, norm=colors.LogNorm())
And that looks like this:
Of course that still has gaps where its value is negative when plotted on a log scale. What your data means to you when the value is negative should dictate how you fill this in. One simple solution is to just set those values to the smallest positive value and they'd fill in as black:
I've plotted a 3-d mesh in Matlab by below little m-file:
[x,n] = meshgrid(0:0.1:20, 1:1:100);
mu = 0;
sigma = sqrt(2)./n;
f = normcdf(x,mu,sigma);
mesh(x,n,f);
I am going to acquire the same result by utilization of Python and its corresponding modules, by below code snippet:
import numpy as np
from scipy.integrate import quad
import matplotlib.pyplot as plt
sigma = 1
def integrand(x, n):
return (n/(2*sigma*np.sqrt(np.pi)))*np.exp(-(n**2*x**2)/(4*sigma**2))
tt = np.linspace(0, 20, 2000)
nn = np.linspace(1, 100, 100)
T = np.zeros([len(tt), len(nn)])
for i,t in enumerate(tt):
for j,n in enumerate(nn):
T[i, j], _ = quad(integrand, -np.inf, t, args=(n,))
x, y = np.mgrid[0:20:0.01, 1:101:1]
plt.pcolormesh(x, y, T)
plt.show()
But the output of the Python is is considerably different with the Matlab one, and as a matter of fact is unacceptable.
I am afraid of wrong utilization of the functions just like linespace, enumerate or mgrid...
Does somebody have any idea about?!...
PS. Unfortunately, I couldn't insert the output plots within this thread...!
Best
..............................
Edit: I changed the linespace and mgrid intervals and replaced plot_surface method... The output is 3d now with the suitable accuracy and smoothness...
From what I see the equivalent solution would be:
import numpy as np
from scipy.stats import norm
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import axes3d
x, n = np.mgrid[0:20:0.01, 1:100:1]
mu = 0
sigma = np.sqrt(2)/n
f = norm.cdf(x, mu, sigma)
fig = plt.figure()
ax = fig.gca(projection='3d')
ax.plot_surface(x, n, f, rstride=x.shape[0]//20, cstride=x.shape[1]//20, alpha=0.3)
plt.show()
Unfortunately 3D plotting with matplotlib is not as straight forward as with matlab.
Here is the plot from this code:
Your Matlab code generate 201 points through x:
[x,n] = meshgrid(0:0.1:20, 1:1:100);
While your Python code generate only 20 points:
tt = np.linspace(0, 19, 20)
Maybe it's causing accuracy problems?
Try this code:
tt = np.linspace(0, 20, 201)
The seminal points to resolve the problem was:
1- Necessity of the equivalence regarding the provided dimensions of the linespace and mgrid functions...
2- Utilization of a mesh with more density to make a bee line into a high degree of smoothness...
3- Application of a 3d plotter function, like plot_surf...
The current code is totally valid...