finding density of points along a line to find the region that has maximum concentration

Question

Finding point with dense neighbourhood

I have a set (count=485) of values in range 0.015 to 0.13 with six digit precision.

I want to use numpy in python .

I have tried

b = []
with open('data.txt') as infile:
    lines = infile.readlines()
    for line in lines:
        for n in line.split()[2:3]:
            b.append(n)
xa = np.asarray(b,dtype=float)
print 'mode-',stats.mode(xa)

Here data.txt has values in third column. This gives value that appears multiple times . In my case 0.06 appeared twice. So, above code does not work for my case.

Can there be a graph using which it can be interpreted that point x has densest neighbourhood if there is a peak at 'x' .

I am not able to define 'neigbourhood' . You can decide by yourself .

Thanks.

Answer 1

you can use matplotlib.pyplot.hist to plot a histogram that will show you a peak. You can also use np.histogram() which will return identical results. Edit: I have used np.argmax() here now on the histogram frequency results to find our largest window. Have also plotted a line on the histogram to show heighest frequency.

You can also check out numpy.genfromtxt() or pandas.read_csv() for easy opening of files.

import matplotlib.pyplot as plt
import numpy as np
#Synthetic Data
dat=np.random.uniform(0.015,0.13,485)#(count=485) of values in range 0.015 to 0.13

binsize=20 #number of windows you want
hist=plt.hist(dat,bins=binsize) #Plot and get our histogram values
#hist=np.histogram(dat,bins=binsize) #will also work, just not plot the answer.

#hist[0] is frequencies and hist[1] is x value

ans=hist[1][np.argmax(hist[0])]
print('Answer: '+str(ans))

buffer=(hist[1][2]-hist[1][1])/2 #Just to centre the black line
plt.axvline(ans+buffer,c='k') #Draw a line of centre
plt.show()