def yt_thetas():
arr = create_ytarray()
X = arr[0]
Y = arr[1]
X = cf.normalize(X)
dimX = X.shape # get dimensions of X as a tuple (rows, columns)
N = dimX[1] # of columns in X; number of features
Theta = np.zeros(N + 1) # add a column for theta0
#Our results
print "Cost before gradient descent: " , cf.calculate_cost(X, Y, Theta)
Results = cf.gradient_descent(X, Y, Theta, .05, 100)
print "Thetas: " , Results[0]
Theta = Results[0]
print "Cost after gradient descent: ", cf.calculate_cost(X, Y, Theta)
return Theta
def main():
#X = np.loadtxt(filename, usecols=(2, 3, 5)) # using displacement, horsepower, and acceleration
#Y = np.loadtxt(filename, usecols=(0,))
#arr = create_ytarray()
arr = create_dmarray()
X = arr[0]
Y = arr[1]
print X
X = cf.normalize(X)
dimX = X.shape # get dimensions of X as a tuple (rows, columns)
N = dimX[1] # of columns in X; number of features
Theta = np.zeros(N + 1) # add a column for theta0
#Our results
print "Cost before gradient descent: " , cf.calculate_cost(X, Y, Theta)
Results = cf.gradient_descent(X, Y, Theta, .005, 2500)
print "Thetas: " , Results[0]
Theta = Results[0]
print "Cost after gradient descent: ", cf.calculate_cost(X, Y, Theta)
# Compare with sci-kit-learns's implementation ##########################################
# Create linear regression object
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(X, Y)
# The coefficients
print "\nResults from sci-kit-learn's linar regression method:"
print 'Coefficients: ', regr.coef_
print 'Intercept : ', regr.intercept_
# The mean square error
# Plot cost versus iterations to check if it converges
pl.xlabel("Iterations")
pl.ylabel("Cost")
pl.plot(Results[1])
pl.show()
def dm_main():
arr = create_dmarray()
X = arr[0]
Y = arr[1]
X = cf.normalize(X)
dimX = X.shape # get dimensions of X as a tuple (rows, columns)
N = dimX[1] # of columns in X; number of features
Theta = np.zeros(N + 1) # add a column for theta0
#Our results
print "Cost before gradient descent: " , cf.calculate_cost(X, Y, Theta)
Results = cf.gradient_descent(X, Y, Theta, .01, 1000)
print "Thetas: " , Results[0]
Theta = Results[0]
print "Cost after gradient descent: ", cf.calculate_cost(X, Y, Theta)
# Compare with sci-kit-learns's implementation ##########################################
# Create linear regression object
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(X, Y)
# The coefficients
print "\nResults from sci-kit-learn's linar regression method:"
print 'Coefficients: ', regr.coef_
print 'Intercept : ', regr.intercept_
# The mean square error
Theta = np.append(regr.coef_, regr.intercept_)
print cf.calculate_cost(X, Y, Theta)
# Plot cost versus iterations to check if it converges
pl.xlabel("Iterations")
pl.ylabel("Cost")
pl.plot(Results[1])
pl.show()
''' X = np.loadtxt(filename, usecols=(2, 3, 5)) # using displacement, horsepower, and acceleration Y = np.loadtxt(filename, usecols=(0,)) X = cf.normalize(X) dimX = X.shape # get dimensions of X as a tuple (rows, columns) N = dimX[1] # of columns in X; number of features Theta = np.zeros(N + 1) # add a column for theta0 #Our results print "Cost before gradient descent: " , cf.calculate_cost(X, Y, Theta) Results = cf.gradient_descent(X, Y, Theta, .01, 500) print "Thetas: " , Results[0] Theta = Results[0] print "Cost after gradient descent: ", cf.calculate_cost(X, Y, Theta) # Compare with sci-kit-learns's implementation ########################################## # Create linear regression object regr = linear_model.LinearRegression() # Train the model using the training sets regr.fit(X, Y) # The coefficients print "\nResults from sci-kit-learn's linar regression method:"
'''
X = np.loadtxt(filename,
usecols=(2, 3,
5)) # using displacement, horsepower, and acceleration
Y = np.loadtxt(filename, usecols=(0, ))
X = cf.normalize(X)
dimX = X.shape # get dimensions of X as a tuple (rows, columns)
N = dimX[1] # of columns in X; number of features
Theta = np.zeros(N + 1) # add a column for theta0
#Our results
print "Cost before gradient descent: ", cf.calculate_cost(X, Y, Theta)
Results = cf.gradient_descent(X, Y, Theta, .01, 500)
print "Thetas: ", Results[0]
Theta = Results[0]
print "Cost after gradient descent: ", cf.calculate_cost(X, Y, Theta)
# Compare with sci-kit-learns's implementation ##########################################
# Create linear regression object
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(X, Y)
# The coefficients
print "\nResults from sci-kit-learn's linar regression method:"
print 'Coefficients: ', regr.coef_
print 'Intercept : ', regr.intercept_