Polynomial fitting in SciPy

Introduction

Polynomial fitting helps us find a smooth curve that best matches a set of points. It is useful to understand trends and make predictions.

You have scattered data points and want to see the overall trend.

You want to predict values between or beyond your data points.

You want to smooth noisy data to see the main pattern.

You want to compare how well different curves fit your data.

Syntax

SciPy

from numpy import polyfit, polyval

coefficients = polyfit(x, y, degree)
fitted_values = polyval(coefficients, x)

polyfit finds the polynomial coefficients that best fit your data.

polyval calculates the y-values using those coefficients.

Examples

Fit a 2nd degree polynomial (a parabola) to points that follow y = x².

SciPy

from numpy import polyfit, polyval

x = [1, 2, 3, 4]
y = [1, 4, 9, 16]
coeffs = polyfit(x, y, 2)
fitted = polyval(coeffs, x)

Fit a straight line (1st degree polynomial) to data points.

SciPy

from numpy import polyfit, polyval

x = [0, 1, 2, 3]
y = [1, 3, 7, 13]
coeffs = polyfit(x, y, 1)
fitted = polyval(coeffs, x)

Sample Program

This code fits a 2nd degree polynomial to some sample data points. It prints the polynomial coefficients and the fitted y-values. Then it shows a plot with the original points and the smooth curve.

SciPy

from numpy import polyfit, polyval
import numpy as np
import matplotlib.pyplot as plt

# Sample data points
x = np.array([0, 1, 2, 3, 4, 5])
y = np.array([2, 3, 5, 10, 18, 30])

# Fit a 2nd degree polynomial
degree = 2
coeffs = polyfit(x, y, degree)

# Calculate fitted values
fitted_y = polyval(coeffs, x)

# Print coefficients
print('Polynomial coefficients:', coeffs)

# Print fitted values
print('Fitted values:', fitted_y)

# Plot original points and fitted curve
plt.scatter(x, y, color='blue', label='Data points')
plt.plot(x, fitted_y, color='red', label='Fitted polynomial')
plt.xlabel('x')
plt.ylabel('y')
plt.title('Polynomial Fitting Example')
plt.legend()
plt.show()

Important Notes

Higher degree polynomials can fit data better but may cause wiggly curves.

Always check if the polynomial degree makes sense for your data.

Plotting helps to see if the fit looks good.

Summary

Polynomial fitting finds a smooth curve to match data points.

Use polyfit to get coefficients and polyval to get fitted values.

Check the fit visually and avoid too high polynomial degrees.

Practice

1. What does the scipy.polyfit function do in polynomial fitting?

easy

A. It calculates the coefficients of the polynomial that best fits the data.

B. It plots the data points on a graph.

C. It predicts future data points without fitting.

D. It normalizes the data before fitting.

Solution

1. Step 1: Understand the purpose of polyfit

   polyfit takes data points and finds polynomial coefficients that best fit those points.

2. Step 2: Differentiate from other functions

   Plotting or normalization are not done by polyfit; it only calculates coefficients.

3. Final Answer:

   It calculates the coefficients of the polynomial that best fits the data. -> Option A

4. Quick Check:

   polyfit = coefficients [OK]

Hint: Remember: polyfit finds coefficients, not plots or predictions [OK]

Common Mistakes:

• Confusing polyfit with plotting functions
• Thinking polyfit predicts future points directly
• Assuming polyfit normalizes data automatically

2. Which of the following is the correct syntax to fit a 3rd degree polynomial to data arrays x and y using SciPy?

easy

A. coeffs = scipy.polyfit(y, x, 3)

B. coeffs = scipy.polyfit(x, y, 3)

C. coeffs = scipy.polyfit(x, y)

D. coeffs = scipy.polyfit(x, y, degree=3)

Solution

1. Step 1: Check the order of arguments in polyfit

   The correct order is polyfit(x, y, degree).

2. Step 2: Confirm the degree argument is positional, not keyword

   polyfit expects degree as the third positional argument, not as a keyword.

3. Final Answer:

   coeffs = scipy.polyfit(x, y, 3) -> Option B

4. Quick Check:

   Correct syntax = coeffs = scipy.polyfit(x, y, 3) [OK]

Hint: Remember: polyfit(x, y, degree) with degree as positional [OK]

Common Mistakes:

• Swapping x and y arguments
• Omitting the degree argument
• Using degree as a keyword argument

3. Given the code:

import numpy as np
from scipy import polyfit, polyval
x = np.array([0, 1, 2, 3])
y = np.array([1, 3, 7, 13])
coeffs = polyfit(x, y, 2)
fitted = polyval(coeffs, x)
print(fitted)

What is the output printed?

medium

A. [ 1. 4. 9. 16.]

B. [ 1. 2. 4. 8.]

C. [ 0. 1. 4. 9.]

D. [ 1. 3. 7. 13.]

Solution

1. Step 1: Fit a 2nd degree polynomial to points

   The points (x, y) fit exactly to y = 1 + 2x + x^2, so polyfit finds coefficients close to [1, 2, 1].

2. Step 2: Use polyval to compute fitted values at x

   Evaluating the polynomial at x gives the original y values: [1, 3, 7, 13].

3. Final Answer:

   [ 1. 3. 7. 13.] -> Option D

4. Quick Check:

   polyval(coeffs, x) = original y [OK]

Hint: polyval with polyfit coeffs returns fitted y values [OK]

Common Mistakes:

• Confusing input arrays order
• Expecting different output than original y
• Misunderstanding polynomial degree effect

4. What is wrong with this code snippet for polynomial fitting?

import numpy as np
from scipy import polyfit, polyval
x = np.array([1, 2, 3])
y = np.array([2, 4, 6])
coeffs = polyfit(x, y, 2)
fitted = polyval(coeffs, x)
print(fitted)

medium

A. The code is correct and will run without errors.

B. The arrays x and y must be lists, not numpy arrays.

C. The degree 2 polynomial is too high for 3 points; use degree 1 instead.

D. polyval cannot be used with coefficients from polyfit.

Solution

1. Step 1: Check polynomial degree vs data points

   Fitting a degree 2 polynomial to 3 points is mathematically valid and will produce a polynomial that fits all points exactly.

2. Step 2: Validate data types and function usage

   Using numpy arrays is correct; polyval works with polyfit coefficients; no syntax errors present.

3. Final Answer:

   The code is correct and will run without errors. -> Option A

4. Quick Check:

   Degree 2 polynomial with 3 points = code runs fine [OK]

Hint: Degree equal to number of points minus one fits exactly [OK]

Common Mistakes:

• Using too high polynomial degree for few points
• Thinking numpy arrays are invalid input
• Believing polyval can't use polyfit output

5. You have noisy data points and want to fit a polynomial that smooths the noise but avoids overfitting. Which approach is best?

hard

A. Use polyfit with degree zero to get a constant fit.

B. Fit a high-degree polynomial to capture all fluctuations.

C. Fit a low-degree polynomial and check the fit visually.

D. Fit multiple polynomials of different degrees and average coefficients.

Solution

1. Step 1: Understand overfitting and noise smoothing

   High-degree polynomials fit noise too closely, causing overfitting; low-degree polynomials smooth data better.

2. Step 2: Use visual check to confirm fit quality

   Plotting fitted curve helps decide if degree is appropriate and avoids overfitting.

3. Final Answer:

   Fit a low-degree polynomial and check the fit visually. -> Option C

4. Quick Check:

   Low degree + visual check = smooth fit [OK]

Hint: Low degree + visual check avoids overfitting [OK]

Common Mistakes:

• Choosing too high degree polynomial
• Using degree zero which ignores trends
• Averaging coefficients from different fits