Python : ZeroDivisionError: float division by zero

Question

I have got these code, But it gives a zero division error. I am not able to figure out what is wrong. I need your help. Thank you. :)

from math import sqrt

def inisialisasi():
    filename = raw_input('File will be read? ')
    return filename


def readdatafile(filename):
    datafile = open('kalibration.txt','r')
    datax = []; datay = []; datae = []; i = 0
    for row in datafile:
        i +=1
        data = row.split()
        x = float(data[0])
        datax.append(x)
        y = float(data[1])
        datay.append(y)
        e = float(data[2])
        datae.append(e)
        print 'x = %5.2f y = %5.2f e = %5.2f' % (x, y, e)

    datafile.close()
    n = i
    print 'Jumlah data = ', n
    return n, datax, datay, datae


def regresilinear(x, y, e):
    sum1=0.0; sum2=0.0; sum3=0.0; sum4=0.0; sum5=0.0
    for i in range(0, n):
        sum1=sum1+(x[i]**2/e[i]**2)
        sum2=sum2+(y[i]/e[i]**2)
        sum3=sum3+(x[i]/e[i]**2)
        sum4=sum4+(x[i]*y[i])/e[i]**2
        sum5=sum5+1/e[i]**2

    det = (sum5*sum1)-sum3**2
    #parameter a dan b
    a = ((sum1*sum2)-(sum3*sum4))/det
    b = ((sum5*sum4)-(sum3)*(sum2))/det
    #ralat
    sigmaa2 = sum1/det
    sigmab2 = sum5/det
    sigmaa = sqrt(sigmaa2)
    sigmab = sqrt(sigmab2)

    return a, b, sigmaa, sigmab

filename = inisialisasi()
n, datax, datay, datae =  readdatafile(filename) 

a, b, sigmaa, sigmab = regresilinear(datax,datay, datae) 

print 'a= %8.6f b= %8.6f sigmaa= %8.6f sigmab= %8.6f' % (a, b, sigmaa, sigmab)

error :

Traceback (most recent call last):

File "coba6.py", line 55, in

a, b, sigmaa, sigmab = regresilinear(datax, datay, datae)

File "coba6.py", line 42, in regresilinear

a = ((sum1*sum2)-(sum3*sum4))/det

ZeroDivisionError: float division by zero

Solution

Somehow, det is set equal to 0. Since you never explicitly prevent that from happening, a single input line like...

1.0 <whatever> 1.0

...can result in division by zero. (The y value doesn't matter in this case.) After substituting, you'll have:

sum1 = 0.0 + 1.0**2 / 1.0**2  # sum1 == 1.0
sum3 = 0.0 + 1.0 / 1.0**2     # sum3 == 1.0
sum5 = 0.0 + 1 / 1.0**2       # sum5 == 1.0
det = 1.0 * 1.0 - 1.0**2      # det  == 0.0
...
a = <whatever> / det  # KABOOM!

Actually, no input will also produce this error, since the for loop in regresilinear will never alter the various sum* variables from their default 0.0 values.

If you're sure your input isn't doing this, you might want to add print statements inside regresilinear to see how det is getting set to zero. (Or use the pdb debugger, if you're familiar with it.)

PS: It would be much easier to debug your functions if they weren't dependent on global data like n. You don't even need it, since it should always equal the length of the three lists datax, datay, and datae.

PPS: readdatafile completely ignores its filename parameter. So if the hard-coded kalibration.txt file happens to be present but empty, you'll get the exact same ZeroDivisionError.