def objectiveFunction(initial_theta):
return lrt.logComputeCost(X, y, initial_theta, lamda)
def objectiveFunction(initial_theta):
return lrt.logComputeCost(X,y,initial_theta, lamda)
print "Binary Classification -- a python implementation of Logistic Regression..."
# Load and normalise data
data = lrt.load_data('log_reg_data.txt')
X_norm, mu, sigma = lrt.normalise_features(data[:, [0, 1]])
X = lrt.addOnes(X_norm)
y = data[:, 2]
# Initialise the model with zeros
theta = np.zeros([1, 3])
# Initialise learning parameters
max_iterations = 400
lamda = 0.1
print "Cost at inital theta (zeros) = ", lrt.logComputeCost(
X, y, theta, lamda)
print "Gradient at inital theta (zeros) = ", lrt.computeGradient(
X, y, theta, lamda)
raw_input("Press Enter to continue...")
#train the model
model = lrt.train(objectiveFunction, theta, max_iterations)
print "model after optimisation", model
print "Cost after optimisation = ", lrt.logComputeCost(X, y, model, lamda)
print "Gradient after optimisation (zeros) = ", lrt.computeGradient(
X, y, model, lamda)
# Create new values for testing the model
new_data = lrt.newDataExample(np.array([45, 85]), mu, sigma)
print "Binary Classification -- a python implementation of Logistic Regression..."
# Load and normalise data
data = lrt.load_data('log_reg_data.txt')
X_norm, mu, sigma = lrt.normalise_features(data[:, [0,1]])
X = lrt.addOnes(X_norm)
y = data[:, 2]
# Initialise the model with zeros
theta = np.zeros([1,3])
# Initialise learning parameters
max_iterations = 400
lamda = 0.1
print "Cost at inital theta (zeros) = ", lrt.logComputeCost(X,y,theta,lamda)
print "Gradient at inital theta (zeros) = ", lrt.computeGradient(X,y,theta, lamda)
raw_input("Press Enter to continue...")
#train the model
model = lrt.train(objectiveFunction, theta, max_iterations)
print "model after optimisation", model
print "Cost after optimisation = ", lrt.logComputeCost(X,y,model,lamda)
print "Gradient after optimisation (zeros) = ", lrt.computeGradient(X,y,model,lamda)
# Create new values for testing the model
new_data = lrt.newDataExample(np.array([45, 85]), mu, sigma)
print "Probability of student passing with 45 as first mark and 85 as second mark = ", lrt.predict(new_data, model)
