numpy - Smoothing in Python -

this question has answer here:

• pandas: rolling mean time interval 7 answers

i use time averaged view data less noisy when plot it. example, if data taken every 1 minute, have 2 arrays, ts , ys. created fs local averaging of 60 nearest points in ys. convolution myself calculating average of 60 nearest points, don't use modules numpy or else.

i have new data ts bit more sparse. is, miss datapoints , can't take average of 60 nearest points. if independent variable, ts, in minutes, how calculate hourly average of dependent variable, ys, create hourly average function, fs, in python?

if independent variable, ts, in minutes, how calculate hourly average of dependent variable, ys, create hourly average function, fs, in python?

this complex problem , possible answers vary depending on mean "hourly average".

one approach dealing irregularly-spaced data resample it. resampling done interpolation, , resulting resampled data usable whatever filter method like.

import numpy np import matplotlib.pyplot plt scipy.signal import savgol_filter %matplotlib inline  def y(t):     # function simulate data     return np.sin(t/20.) + 0.05*np.random.randn(len(t))  four_hours = np.arange(240) random_time_points = np.sort(np.random.choice(four_hours, size=30, replace=false))  simulated_data = y(random_time_points) resampled_data = np.interp(four_hours, random_time_points, simulated_data)  # here smooth savitzky-golay filter,  #  use moving avg or else #  window-length=61 means smooth on 1-hour (60 minute) window smoothed_data = savgol_filter(resampled_data, window_length=61, polyorder=0)  # plot results plt.plot(random_time_points, simulated_data, '.k',           four_hours, smoothed_data, '--b',          four_hours, y(four_hours), '-g')  # save plot plt.savefig('so35038933.png') 

the plot shows original "sparse" data (black points), original "true" data (green curve), , smoothed data (blue dotted curve).