9

I have a question.

I have plotted a graph using Matplotlib like this:

from matplotlib import pyplot
import numpy
from scipy.interpolate import spline

widths = numpy.array([0, 30, 60, 90, 120, 150, 180])
heights = numpy.array([26, 38.5, 59.5, 82.5, 120.5, 182.5, 319.5])

xnew = numpy.linspace(widths.min(),widths.max(),300)
heights_smooth = spline(widths,heights,xnew)

pyplot.plot(xnew,heights_smooth)
pyplot.show()

Now I want to query a height value using width value as an argument. I cannot seem to find how to do that. Please help! Thanks in advance!

Subhamoy S.
  • 6,566
  • 10
  • 37
  • 53

3 Answers3

9

plot() returns a useful object: [<matplotlib.lines.Line2D object at 0x38c9910>]
From that we can get x- and y-axis values: 

import matplotlib.pyplot as plt, numpy as np
...
line2d = plt.plot(xnew,heights_smooth)
xvalues = line2d[0].get_xdata()
yvalues = line2d[0].get_ydata()

Then we can get the index of one of the width values:

idx = np.where(xvalues==xvalues[-2]) # this is 179.3979933110368
# idx is a tuple of array(s) containing index where value was found
# in this case -> (array([298]),)

And the corresponding height:

yvalues[idx]
# -> array([ 315.53469])

To check we can use get_xydata():

>>> xy = line2d[0].get_xydata()
>>> xy[-2]
array([ 179.39799331,  315.53469   ])
mechanical_meat
  • 163,903
  • 24
  • 228
  • 223
  • 2
    Thank you for the reply! I had to make only one minor adjustment to it: `idx=(numpy.abs(xvalues-)).argmin()`. After this, `yvalues[idx]` gives me what I want. :) – Subhamoy S. Mar 24 '12 at 13:22
0

Here's another option if you're willing to use a different spline function:

from matplotlib import pyplot
import numpy
from scipy import interpolate

widths = numpy.array([0, 30, 60, 90, 120, 150, 180])
heights = numpy.array([26, 38.5, 59.5, 82.5, 120.5, 182.5, 319.5])

xnew = numpy.linspace(widths.min(),widths.max(),300)
heights_smooth = interpolate.splrep(widths,heights) #Use splrep instead of spline

#Select desired width values
width_vals = [0, 80.5, 38.98743]   

#splev returns the value of your spline evaluated at the width values.    
heights = interpolate.splev(width_vals, heights_smooth)

Then

In[]:  heights
Out[]: array([ 26.        ,  74.1721985 ,  44.47929453])

Or evaluate at a point:

w = 167.2
heights = interpolate.splev(w, heights_smooth)
height = heights.item()

In[]:  height
Out[]: 247.8396196684303

The .item() function is necessary because splev returns an array()

Jacob Stern
  • 3,758
  • 3
  • 32
  • 54
0

You could cast the array to a list:

>>> heights[list(widths).index(30)]
38.5

for the interpolated result:

s = xnew[56] 
print s, heights_smooth[list(xnew).index(s)]
33.7123745819, 40.9547542163

As xnew is an ordered list you could use the bisect module to find a closest width value for a queried width, and then find the corresponding height, in a similar fashion:

....
import bisect
pyplot.plot(xnew,heights_smooth)
#33.1222 is a queried value which does not exist in xnew.
index_of_nearest_width = bisect.bisect_left(xnew, 33.1222) 
width_val = xnew[index_of_closest_width]
print width_val, heights_smooth[list(xnew).index(width_val)]
#prints the nearest width to 33.1222 then the corresponding height.
33.7123745819 40.9547542163
fraxel
  • 34,470
  • 11
  • 98
  • 102