I am creating an n-sized polynomial, which I want to fit over my data. Because of the decorator it is supposed to print "test" every time the function is run, however, because of the decorator, the curve_fit no longer knows how many parameters it should supply (I get the error: ValueError: Unable to determine number of fit parameters.). Is there a way to make the curve_fit "know" how many parameters it should supply?
import numpy as np
import matplotlib.pyplot as plt
from scipy.optimize import curve_fit
def print_dec(func_): # prints test before function
def inner(*args, **kwargs):
print("test")
returned_value = func_(*args, **kwargs)
return returned_value
return inner
n = 10
func_str = "lambda x" # function creation
for i in range(n+1):
func_str += f", var_{i}"
func_str += ": var0"
for i in range(1, n+1):
func_str += f" + var_{i}*x**{i}"
print(func_str)
func = eval(func_str)
func = print_dec(func)
x_data = np.linspace(0, 20, 1000)
y_data = x_data + np.random.random(1000) # data is just a straight line with noise
p_opt, p_cov = curve_fit(func, xdata=x_data, ydata=y_data)
print(p_opt) # simple curve_fit
plt.plot(x_data, func(x_data, *p_opt), c='r') # fitted curve
plt.scatter(x_data, y_data) # original data
plt.show()
Your function is dynamically capturing parameters with your *args and **kwargs, but curve_fit needs to know how many parameters your function is using. You can pass p0=[1]*(n+1) in your case to solve the error. You can review the curve_fit implementation/documentation for more details.
You'll also need to fix the line: func_str += ": var_0" (adding an underscore on your var0 as expected on the lambda function).
Also, on the implementation, you could use a loop instead of dynamically creating a lambda function, it would be easier to debug, and probably easier to read too