Python:Fitting

From PrattWiki
Revision as of 00:54, 7 November 2018 by DukeEgr93 (talk | contribs)
Jump to navigation Jump to search

This document contains examples of polynomial fitting, general linear regression, and nonlinear regression. In each section, there will be example code that may come in useful for later courses. The example code is based on the existence of a file in the same directory called Cantilever.dat that contains two columns of data - the first is an amount of mass (in kg) placed at the end of a beam and the second is a displacement, measured in inches, at the end of the beam. For EGR 103, this file is:

0.000000        0.005211
0.113510002     0.158707
0.227279999     0.31399
0.340790009     0.474619
0.455809998     0.636769
0.569320007     0.77989
0.683630005     0.936634
0.797140015     0.999986

Common Command Reference

All links below to NumPy v1.15 manual at NumPy v1.15 Manual; these commands show up in just about all the examples:

In the scripts below, common code has a regular background while code that differs from script to script will be highlighted in yellow.

Polynomial Fitting

Polynomial fits are those where the dependent data is related to some set of integer powers of the independent variable. MATLAB's built-in polyfit command can determine the coefficients of a polynomial fit.

Specific Command References

All links below to NumPy v1.15 manual at NumPy v1.15 Manual

Example Code

In the example code below, n determines the order of the fit. Not much else would ever need to change.

 1 # %% Import modules
 2 import numpy as np
 3 import matplotlib.pyplot as plt
 4 
 5 
 6 # %% Load and manipulate data
 7 # Load data from Cantilever.dat
 8 beam_data = np.loadtxt('Cantilever.dat')
 9 # Copy data from each column into new variables
10 mass = beam_data[:, 0].copy()
11 disp = beam_data[:, 1].copy()
12 # Convert mass to force
13 force = mass * 9.81
14 # Convert disp to meters
15 disp = (disp * 2.54) / 100
16 
17 # %% Rename and create model data
18 x = force
19 y = disp
20 xmodel = np.linspace(np.min(x), np.max(x), 100)
21 
22 # %% Perform calculations
23 n = 1
24 p = np.polyfit(x, y, n)
25 print(p)
26 
27 
28 
29 
30 # %% Generate estimates and model
31 yhat = np.polyval(p, x)
32 ymodel = np.polyval(p, xmodel)
33 
34 # %% Calculate statistics
35 st = np.sum((y - np.mean(y))**2)
36 sr = np.sum((y - yhat)**2)
37 r2 = (st - sr) / st
38 print('st: {}\nsr: {}\nr2: {}'.format(st, sr, r2))
39 
40 # %% Generate and save plots
41 plt.figure(1)
42 plt.clf()
43 plt.plot(x, y, 'ko', label='Data')
44 plt.plot(x, yhat, 'ks', label='Estimates', mfc='none')
45 plt.plot(xmodel, ymodel, 'k-', label='Model')
46 plt.grid(1)
47 plt.legend()

General Linear Regression

General linear regression involves finding some set of coefficients for fits that can be written as:

\( \hat{y}(x)=\sum_{j=1}^{M}c_j\phi_j(x) \)

where the \(c_j\) are the coefficients of the fit and the \(\phi_j\) are the specific functions of the independent variable that make up the fit.

Specific Command References

All links below to NumPy v1.15 manual at NumPy v1.15 Manual

Example Code

In the example code below, there is an example of a general linear fits of one variable. It is solving the same fit as given above, just in different way. Specifically it uses linear algebra to find the coefficients that minimize the sum of the squares of the estimate residuals for a general linear fit. In this code, variables ending in "v" explicitly need to be column vectors while variables ending in "e" can either be 1D arrays or 2D arrays (or lists).

 1 # %% Import modules
 2 import numpy as np
 3 import matplotlib.pyplot as plt
 4 
 5 
 6 # %% Load and manipulate data
 7 # Load data from Cantilever.dat
 8 beam_data = np.loadtxt('Cantilever.dat')
 9 # Copy data from each column into new variables
10 mass = beam_data[:, 0].copy()
11 disp = beam_data[:, 1].copy()
12 # Convert mass to force
13 force = mass * 9.81
14 # Convert disp to meters
15 disp = (disp * 2.54) / 100
16 
17 # %% Rename and create model data
18 xv = np.reshape(force, (-1, 1))
19 yv = np.reshape(disp, (-1, 1))
20 xmodel = np.linspace(np.min(xv), np.max(xv), 100)
21 
22 # %% Perform calculations
23 def yfun(xe, coefs):
24     return coefs[0] * xe**1 + coefs[1] * xe**0
25 
26 a_mat = np.block([[xv**1, xv**0]])
27 pvec = np.linalg.lstsq(a_mat, yv, rcond=None)[0]
28 print(pvec)
29 
30 # %% Generate estimates and model
31 yhat = yfun(xv, pvec)
32 ymodel = yfun(xmodel, pvec)
33 
34 # %% Calculate statistics
35 st = np.sum((yv - np.mean(yv))**2)
36 sr = np.sum((yv - yhat)**2)
37 r2 = (st - sr) / st
38 print('st: {}\nsr: {}\nr2: {}'.format(st, sr, r2))
39 
40 # %% Generate and save plots
41 plt.figure(1)
42 plt.clf()
43 plt.plot(xv, yv, 'ko', label='Data')
44 plt.plot(xv, yhat, 'ks', label='Estimates', mfc='none')
45 plt.plot(xmodel, ymodel, 'k-', label='Model')
46 plt.grid(1)
47 plt.legend()

Nonlinear Regression

Nonlinear regression is both more powerful and more sensitive than linear regression. For inherently nonlinear fits, it will also produce a better \(S_r\) value than linearization since the nonlinear regression process is minimizing the \(S_r\) of the actual data rather than that of the transformed values. The sensitivity comes into play as the optimization routine may find local minima versus global minima. A good starting guess will work wonders.

Specific Command References

The link below is to the SciPy v1.1.0 reference guide at SciPy

Example Code

Note in the example code that the initial guess gives 0.6 for the slope and 0.1 for the intercept. While these numbers are quite far from the optimized values of 0.0034 for the slope and 0.00055 for the intercept, the optimization routine is still able to find the correct value. That is not always the case - try to find an initial guess close to the actual answer.

 1 # %% Import modules
 2 import numpy as np
 3 import matplotlib.pyplot as plt
 4 import scipy.optimize as opt
 5 
 6 # %% Load and manipulate data
 7 # Load data from Cantilever.dat
 8 beam_data = np.loadtxt('Cantilever.dat')
 9 # Copy data from each column into new variables
10 mass = beam_data[:, 0].copy()
11 disp = beam_data[:, 1].copy()
12 # Convert mass to force
13 force = mass * 9.81
14 # Convert disp to meters
15 disp = (disp * 2.54) / 100
16 
17 # %% Rename and create model data
18 x = force
19 y = disp
20 xmodel = np.linspace(np.min(x), np.max(x), 100)
21 
22 # %% Perform calculations
23 def yfun(x, *coefs):
24     return coefs[0] * x**1 + coefs[1] * x**0
25 
26 popt = opt.curve_fit(yfun, x, y, [0.6, 0.1])[0]
27 print(popt)
28 
29 
30 # %% Generate estimates and model
31 yhat = yfun(x, *popt)
32 ymodel = yfun(xmodel, *popt)
33 
34 # %% Calculate statistics
35 st = np.sum((y - np.mean(y))**2)
36 sr = np.sum((y - yhat)**2)
37 r2 = (st - sr) / st
38 print('st: {}\nsr: {}\nr2: {}'.format(st, sr, r2))
39 
40 # %% Generate and save plots
41 plt.figure(1)
42 plt.clf()
43 plt.plot(x, y, 'ko', label='Data')
44 plt.plot(x, yhat, 'ks', label='Estimates', mfc='none')
45 plt.plot(xmodel, ymodel, 'k-', label='Model')
46 plt.grid(1)
47 plt.legend()

Example changes for different models

Polynomial

For the polynomial fitting model, really the only thing that would change would be the order of the fit and thus the value of \(n\) on line 23 of that code.

General Linear

For the general linear fit, the two places things will change will be in the function definition on line 24 and in the creation of the block matrix on line 26; the changes will mirror each other. For the straight line model - formally:

\( \hat{y}(x)=mx^1+bx^0 \)

the code is

22 # %% Perform calculations
23 def yfun(xe, coefs):
24     return coefs[0] * xe**1 + coefs[1] * xe**0
25 
26 a_mat = np.block([[xv**1, xv**0]])
27 pvec = np.linalg.lstsq(a_mat, yv, rcond=None)[0]
28 print(pvec)

where pvec[0][0] will be the slope \(m\) and pvec[1][0] will be the intercept \(b\). If the model were changed to, say,

\( \hat{y}(x)=a\cos(x)+b\sin(x)+c \)

which, formally, is

\( \hat{y}(x)=a\cos(x)+b\sin(x)+cx^0 \)

then the code would change to:

22 # %% Perform calculations
23 def yfun(xe, coefs):
24     return coefs[0] * np.cos(xe) + coefs[1] * np.sin(xe) + coefs[2] * xe**0
25 
26 a_mat = np.block([[np.cos(xv), np.sin(xv), xv**0]])
27 pvec = np.linalg.lstsq(a_mat, yv, rcond=None)[0]
28 print(pvec)

and pvec will be a 2D array with three rows.

Nonlinear

For the nonlinear fit, the two places things will change will be in the function definition on line 24 and in the initial parameter value guess on line ; there need to be as many values in the initial parameter list as there are parameters. For the straight line model - formally:

\( \hat{y}(x)=mx^1+bx^0 \)

the code is

22 # %% Perform calculations
23 def yfun(x, *coefs):
24     return coefs[0] * x**1 + coefs[1] * x**0
25 
26 popt = opt.curve_fit(yfun, x, y, [0.6, 0.1])[0]
27 print(popt)

where popt[0] will be the slope \(m\) and popt[1] will be the intercept \(b\). If the model were changed to, say,

\( \hat{y}(x)=a\cos(bx+c)+d \)

which, formally, is

\( \hat{y}(x)=a\cos(bx+c)+dx^0 \)

then the code would change to:

22 # %% Perform calculations
23 def yfun(x, *coefs):
24     return coefs[0] * np.cos(coefs[1]*x + coefs[2]) + coefs[3] * x**0
25 
26 popt = opt.curve_fit(yfun, x, y, [.01, 1, np.pi/4, .0125])[0]
27 print(popt)

where [.01, 1, np.pi/4, .0125] represents initial guesses for the magnitude a, frequency b, phase c, and average value of the function. The popt variable would have 4 values.

References