Not getting a proper fit

Question

Getting an error of "IndexError: index 1 is out of bounds for axis 0 with size 1". I am a newbie. Please help. Thanks in advance.

def logistic(x, l, k, x1):
    return l / 1+np.exp(-k*(x-x1))
distance= [1.000*70, 2.000*70, 3.000*70, 4.000*70, 5.000*70, 6.000*70, 7.000*70, 8.000*70,
           9.000*70, 11.00*70, 12.000*70, 13.000*70, 14.000*70, 15.000*70, 16.000*70,
           17.000*70, 18.000*70, 19.000*70, 21.000*70, 22.000*70, 23.000*70, 24.000*70, 25.000*70, 26.000*70,
           27.000*70, 28.000*70, 29.000*70, 30.000*70, 31.000*70, 32.000*70, 33.000*70,
           34.000*70, 35.000*70, 36.000*70]
amplitude= [26, 31, 29, 26, 27, 24, 24, 28, 24, 24, 28, 31, 24, 26, 55, 30, 73, 101, 168, 219, 448, 833, 1280, 1397, 1181, 1311,
            1715, 1975, 2003, 2034, 2178, 2180, 2182]
plt.plot(distance,amplitude, 'o')
popt, pcov = curve_fit(logistic, distance, amplitude,maxfev=100, bounds=((100, 10, 0), (200000, 200000, 200000)),p0=[2700, 3000, 1200])
print(popt)
plt.plot(distance, logistic(distance, *popt), 'r', label='logistic fit')
plt.show()

Asked By: Ben Jack

||

Source

Answer 1

there are few points on your code which had to be fixed.

distance and amplititle are not same size
both are list however inside you logistics definition you treat them as numpy array which you do vector operations
bounds value for k is wrong, it should be below 1 but you set minimum value to be 10, which make optimizing difficult
logistic definition you could add some safety measure for stability

from scipy.optimize import curve_fit

from scipy.optimize import curve_fit 
def logistic(x, l, k, x1):
    e =  -k * (x - x1)
    e[ e > 30 ] = 30
    return l / (1 + np.exp(e))

distance= [1.000*70, 2.000*70, 3.000*70, 4.000*70, 5.000*70, 6.000*70, 7.000*70, 8.000*70,
           9.000*70, 11.00*70, 12.000*70, 13.000*70, 14.000*70, 15.000*70, 16.000*70,
           17.000*70, 18.000*70, 19.000*70, 21.000*70, 22.000*70, 23.000*70, 24.000*70, 25.000*70, 26.000*70,
           27.000*70, 28.000*70, 29.000*70, 30.000*70, 31.000*70, 32.000*70, 33.000*70,
           34.000*70, 35.000*70, 36.000*70]
amplitude= [26, 31, 29, 26, 27, 24, 24, 28, 24, 24, 28, 31, 24, 26, 55, 30, 73, 101, 168, 219, 448, 833, 1280, 1397, 1181, 1311,
            1715, 1975, 2003, 2034, 2178, 2180, 2182]
plt.plot(distance[1:], amplitude, 'o')
popt, pcov = curve_fit(logistic, np.array(distance[1:]), np.array(amplitude[:]),maxfev=100, bounds=((100, 0, 0), (3000, 1, 200000)),p0=[2700, 0.1, 1200])

print(popt)
plt.plot(distance, logistic(distance, *popt), 'r', label='logistic fit')

plt.show()

output:

Answered By: amirhm

Not getting a proper fit

Question:

Answers: