I was working on “Machine learning in Robotics” , unit 2, i was trying to run the program under " 2.2 Linear regression - gradient descent", when i tried to run the program, i was getting the error shown in the screenshot. Can you tell me how to fix this?
Does it have any connection to an already reported error:
I think i solved it, please add #! /usr/bin/env python to the program (not provided in the program provided) and run the command pip install future
I furthur got the following errors:
/home/user/catkin_ws/src/machine_learning_course/unit2/src/gradient_decent.py:26: RuntimeWarning: overflow encountered in double_scalars
totalError += (y[i]-(m*x[i] + b)) ** 2
/home/user/catkin_ws/src/machine_learning_course/unit2/src/gradient_decent.py:55: RuntimeWarning: overflow encountered in double_scalars
gradient_m = (-2/n) * sum(X * (Y - Y_pred))
/home/user/catkin_ws/src/machine_learning_course/unit2/src/gradient_decent.py:59: RuntimeWarning: invalid value encountered in double_scalars
m = m - alpha * gradient_m
The program is as below:
#! /usr/bin/env python
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import future
MSE = []
iteration = []
data = pd.read_csv('/home/user/catkin_ws/src/machine_learning_course/dataset/test_brakes.csv')
data.drop(['Unnamed: 0'], axis=1, inplace=True)
data.drop(['payload'], axis=1, inplace=True)
X = (data.iloc[:,1].values).flatten()
Y = (data.iloc[:,:-1].values).flatten()
def compute_error_for_line_given_points(b,m,X,Y):
totalError = 0 #sum of square error formula
for i in range (0, len(X)):
x = X
y = Y
totalError += (y[i]-(m*x[i] + b)) ** 2
mse = (totalError)/totalError.size
return mse
#MSE.append(mse)
#iteration.append(i)
#print (MSE)
#return totalError/ float(len(X))
# Initial values
m = 0
b = 0
#Hyperparameters
alpha = 0.0001 # The learning Rate
epochs = 15000 # The number of iterations to perform gradient descent
#n = float(len(X))
n = len(X)
##########################################################################
# OUR OPTIMIZATOR - we use MSE to look for the best paramaters m and b
# We use Gradient Descent
#########################################################################
for i in range(epochs):
Y_pred = m*X + b
#IMPLEMENTATION OF DISCUSSED ABOVE EQATIONS FOR GRADIENT DESCENT
gradient_m = (-2/n) * sum(X * (Y - Y_pred))
gradient_b = (-2/n) * sum(Y - Y_pred)
#UPDATE OF m AND b
m = m - alpha * gradient_m
b = b - alpha * gradient_b
mse = compute_error_for_line_given_points(b,m,X,Y)
MSE.append(mse)
iteration.append(i)
# performing the regression baseon od gradient descent
Y_pred = m*X + b
plt.figure(figsize=(10, 8))
plt.scatter(X, Y)
plt.plot([min(X), max(X)], [min(Y_pred), max(Y_pred)], color='red') # predicted
plt.xlabel("% of max speed of axis 1", fontsize=16)
plt.ylabel("stop distance [deg]", fontsize=16)
plt.title('Gradient descent', fontsize=18)
plt.show()
plt.figure(figsize=(10, 8))
i = np.arange(0, len(MSE),1)
plt.xlabel("epoch", fontsize=16)
plt.ylabel("error", fontsize=16)
plt.title('Gradient descent ERROR (MSE)', fontsize=18)
plt.plot(i,MSE)
plt.show()
You can ignore the warning unless it’s stopping the program from working properly.
the web shell is stuck with the error as shared. I am not able to proceed with the program.
This seems to be because “the result may be too large to be expressed”.
Does this problem happen whenever you run the script? If so, we definitely have to solve it from our side.
I ran the program as is with only the addition of the #! /usr/bin/env python line at the top. I also had to install the future using pip install future.
Please look into it, do let me know if you need anymore info