Alvaro Hernandorena - 5 months ago 88
Python Question

# Can I make a logarithmic regression on sklearn?

I don't know if "logarithmic regression" is the right term, I need to fit a curve on my data, like a polynomial curve but going flat on the end.

Here is an image, the blue curve is what I have (2nd order polynomial regression) and the magenta curve is what I need.

I have search a lot and can't find that, only linear regression, polynomial regression, but no logarithmic regression on sklearn. I need to plot the curve and then make predictions with that regression.

EDIT

Here is the data for the plot image that I posted:

``````x,y
670,75
707,46
565,47
342,77
433,73
472,46
569,52
611,60
616,63
493,67
572,11
745,12
483,75
637,75
218,251
444,72
305,75
746,64
444,98
342,117
272,85
128,275
500,75
654,65
241,150
217,150
426,131
155,153
841,66
737,70
722,70
754,60
664,60
688,60
796,55
799,62
229,150
232,95
116,480
340,49
501,65
``````

If I understand correctly, you want to fit the data with a function like y = a * exp(-b * (x - c)) + d.

I am not sure if sklearn can do it. But you can use scipy.optimize.curve_fit() to fit your data with whatever the function you define.(scipy):

For your case, I experimented with your data and here is the result:

``````import numpy as np
import matplotlib.pyplot as plt
from scipy.optimize import curve_fit

my_data = np.genfromtxt('yourdata.csv', delimiter=',')
my_data = my_data[my_data[:,0].argsort()]
xdata = my_data[:,0].transpose()
ydata = my_data[:,1].transpose()

# define a function for fitting
def func(x, a, b, c, d):
return a * np.exp(-b * (x - c)) + d

init_vals = [50,0,90,63]
# fit your data and getting fit parameters
popt, pcov = curve_fit(func, xdata, ydata, p0=init_vals, bounds=.   ([0,0,90,0], [1000, 0.1, 200, 200]))
# predict new data based on your fit
y_pred = func(200, *popt)
print(y_pred)

plt.plot(xdata, ydata, 'bo', label='data')
plt.plot(xdata, func(xdata, *popt), '-', label='fit')
plt.xlabel('x')
plt.ylabel('y')
plt.legend()
plt.show()
``````

I found that the initial value for `b` is critical for fitting. I estimated a small range for it and then fit the data.

If you have no priori knowledge of the relationship between `x` and `y`, you can use the regression methods provided by sklearn, like linear regression, Kernel ridge regression (KRR), Nearest Neighbors Regression, Gaussian Process Regression etc. to fit nonlinear data. Find the documentation here

Recommended from our users: Dynamic Network Monitoring from WhatsUp Gold from IPSwitch. Free Download