In 

[138…
#1. Test Points

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

#generate 15 random points using uniform distribution

x_list = rng.uniform(-2, 2, 15)

x_list = np.sort(x_list, 0)

y_list = []

#generate y values

for i in x_list:

y = i**3 + 0.5 * np.random.normal(0,1,1)

y_list = np.append(y_list, y, axis=0)

#convert x and y to dataframe

df = pd.DataFrame(np.column_stack([x_list, y_list]), columns=['x','y'])

print(df)

x y

0 -1.616408 -5.322508

1 -1.253063 -1.743096

2 -1.090656 -2.131412

3 -1.007017 -1.976215

4 -0.732967 0.004277

5 -0.668744 -0.161225

6 -0.435562 0.208111

7 -0.232641 0.532104

8 0.393235 0.006126

9 0.668950 0.516830

10 0.691024 1.021336

11 0.705019 0.836321

12 1.189462 1.788390

13 1.767211 5.752160

14 1.795525 5.621833

In [139…
#Linear Regression

from sklearn.linear_model import LinearRegression

#Linear Regression object

lin_reg = LinearRegression()

x_list = x_list.reshape(-1, 1)

y_list = y_list.reshape(-1, 1)

#Line of best fit

lin_reg.fit(x_list, y_list)

plt.scatter(x_list, y_list, color = 'red')

plt.plot(x_list, lin_reg.predict(x_list))

plt.title('Linear Regression')

plt.xlabel('X values')

plt.ylabel('Y values')

plt.show()

In [150…
#Polynomial Regression with Degree 2

from sklearn.preprocessing import PolynomialFeatures

poly_reg = PolynomialFeatures(degree = 2)

x_poly_2 = poly_reg.fit_transform(x_list)

linreg = LinearRegression()

linreg.fit(x_poly_2, y_list)

#Line of best fit

lin_reg.fit(x_list, y_list)

plt.scatter(x_list, y_list, color = 'red')

plt.plot(x_list, linreg.predict(x_poly_2))

plt.title('Polynomial Regression Model with Degree-2')

plt.xlabel('X values')

plt.ylabel('Y values')

plt.show()

In [151…
#Polynomial Regression with Degree 3

from sklearn.preprocessing import PolynomialFeatures

poly_reg = PolynomialFeatures(degree = 3)

x_poly_3 = poly_reg.fit_transform(x_list)

linreg = LinearRegression()

linreg.fit(x_poly_3, y_list)

#Line of best fit

lin_reg.fit(x_list, y_list)

plt.scatter(x_list, y_list, color = 'red')

plt.plot(x_list, linreg.predict(x_poly_3))

plt.title('Polynomial Regression Model with Degree-3')

plt.xlabel('X values')

plt.ylabel('Y values')

plt.show()

In [152…
#Polynomial Regression with Degree 4

from sklearn.preprocessing import PolynomialFeatures

poly_reg = PolynomialFeatures(degree = 4)

x_poly_4 = poly_reg.fit_transform(x_list)

linreg = LinearRegression()

linreg.fit(x_poly_4, y_list)

#Line of best fit

lin_reg.fit(x_list, y_list)

plt.scatter(x_list, y_list, color = 'red')

plt.plot(x_list, linreg.predict(x_poly_4))

plt.title('Polynomial Regression Model with Degree-4')

plt.xlabel('X values')

plt.ylabel('Y values')

plt.show()

In [184…
#Combine all models

fig, ax = plt.subplots()

ax = fig.add_subplot(1,1,1)

ax.scatter(x_list, y_list)

ax.plot(x_list, lin_reg.predict(x_list))

plot_color = ['blue', 'orange', 'yellow', 'green']

j = 0

for i in range(1, 5):

poly = PolynomialFeatures(degree = i)

x_poly_i = poly_reg.fit_transform(x_list)

linear_reg = LinearRegression()

linear_reg.fit(x_poly_i, y_list)

ax.plot(x_list, linreg.predict(x_poly_i), color = plot_color[j])

j += 1

ax.set_title('Regression Models')
ax.legend(['quatric', 'linear', 'quadratic', 'cubic'],loc = 'upper left')

ax.set_xlabel('X values')

ax.set_ylabel('Y values')

plt.show()

In [166…
#Test the models

rng = np.random.default_rng(12345)

 x1 = rng.uniform(-2, 2, 100)

x1 = np.sort(x1, 0)

y1 = []

#generate y values

for i in x1:

y = i**3 + 0.5 * np.random.normal(0,1,1)

y1 = np.append(y1, y, axis=0)

#convert x and y to dataframe

df = pd.DataFrame(np.column_stack([x1, y1]), columns=['x','y'])

print(df.head(10))

x y

0 -1.979911 -8.547646

1 -1.961151 -7.002029

2 -1.767190 -5.984456

3 -1.727649 -4.418418

4 -1.676513 -4.710488

5 -1.673622 -5.103340

6 -1.657713 -5.133752

7 -1.633341 -4.693403

8 -1.616408 -5.052430

9 -1.482124 -3.340472

In [168…
plt.scatter(x1, y1)

plt.title('Scatter Plot')

plt.xlabel('X values')

plt.ylabel('Y values')

plt.show()

In [177…
#Root Mean Squared Error

from sklearn.metrics import mean_squared_error

lin_reg = LinearRegression()

lin_reg.fit(x_list, y_list)

y_poly_pred_1 = lin_reg.predict(x_list)

rmse1 = np.sqrt(mean_squared_error(y_list,y_poly_pred_1))

print("Root Mean Squared Error for Degree 1:", rmse1)

poly_reg = PolynomialFeatures(degree = 2)

x_poly_2 = poly_reg.fit_transform(x_list)

linreg.fit(x_poly_2, y_list)

 y_poly_pred_2 = linreg.predict(x_poly_2)

rmse2 = np.sqrt(mean_squared_error(y_list,y_poly_pred_2))

print("Root Mean Squared Error for Degree 2:", rmse2)

poly_reg = PolynomialFeatures(degree = 3)

x_poly_3 = poly_reg.fit_transform(x_list)

linreg.fit(x_poly_3, y_list)

y_poly_pred_3 = linreg.predict(x_poly_3)

rmse3 = np.sqrt(mean_squared_error(y_list,y_poly_pred_3))

print("Root Mean Squared Error for Degree 3:", rmse3)

poly_reg = PolynomialFeatures(degree = 4)

x_poly_4 = poly_reg.fit_transform(x_list)

linreg.fit(x_poly_4, y_list)

y_poly_pred_4 = linreg.predict(x_poly_4)

rmse4 = np.sqrt(mean_squared_error(y_list,y_poly_pred_4))

print("Root Mean Squared Error for Degree 4:", rmse4)

Root Mean Squared Error for Degree 1: 1.1214415839339449

Root Mean Squared Error for Degree 2: 1.1083379842942511

Root Mean Squared Error for Degree 3: 0.3649095497100184

Root Mean Squared Error for Degree 4: 0.3632191621497789

In [ ]:
#Polynomial with Degree 4 has the lowest RMS error. So, it's the best model among al