#L1-regularized LS(LASSO)
LASSO_theta=partOne.LASSO_Theta(aftermat_sampleX,sampleY,5)
LASSO_result=partOne.LASSO_predi(aftermat_polyx, LASSO_theta)
LASSO_error = partOne.error(LASSO_result, polyy)
print('LASSO_error:', LASSO_error)
plt.figure(3)
plt.title('L1-regularized LS(LASSO)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, LASSO_result, '.-')
plt.plot(sampleX, sampleY, '*')
#Robust regression(RR)
RR_theta=partOne.RR_Theta(sampleX,sampleY,aftermat_sampleX,korder)
# print(RR_theta.shape)
RR_result=partOne.RR_predi(aftermat_polyx, RR_theta)
RR_error = partOne.error(RR_result, polyy)
print('RR_error:', RR_error)
plt.figure(4)
plt.title('Robust regression(RR)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, RR_result, '.-')
plt.plot(sampleX, sampleY, '*')
# plt.show()
# Bayesian regression (BR)
BR_cov, BR_mean = partOne.posterior_BR(aftermat_sampleX, sampleY)
average, fangcha = partOne.BR_predi(aftermat_polyx, BR_cov, BR_mean)
BR_polyy = np.random.normal(average, np.sqrt(np.abs(fangcha)), size=None)
# print('average.shape:', average.shape)
def xunhun(per):
[sampleX1, sampleY1, polyx, polyy] = partOne.acquire_date()
num_size = round(sampleX1.size * per)
location = random.sample(range(1, sampleX1.size), num_size)
selection_sampleX = []
selection_sampleY = []
for loca in location:
selection_sampleX.append(sampleX1[loca])
selection_sampleY.append(sampleY1[loca])
sampleX = np.array(selection_sampleX).reshape(len(selection_sampleX), 1)
sampleY = np.array(selection_sampleY).reshape(len(selection_sampleY), 1)
korder = 5
# least_squares(LS)
aftermat_sampleX = partOne.theta_mat(sampleX, korder)
LS_theta = partOne.least_squ_theta(aftermat_sampleX, sampleY)
aftermat_polyx = partOne.theta_mat(polyx, korder)
LS_result = partOne.least_squ_predi(aftermat_polyx, LS_theta)
LS_error = partOne.error(LS_result, polyy)
# print('LS_error:', LS_error)
plt.figure(1)
plt.title('least_squares(LS)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, LS_result, '.-')
plt.plot(sampleX, sampleY, '*')
# Regularized LS(RLS)
RLS_theta = partOne.RLS_theta(aftermat_sampleX, sampleY)
RLS_result = partOne.RLS_predi(aftermat_polyx, RLS_theta)
RLS_error = partOne.error(RLS_result, polyy)
# print('RLS_error:', RLS_error)
plt.figure(2)
plt.title('Regularized LS(RLS)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, RLS_result, '.-')
plt.plot(sampleX, sampleY, '*')
# L1-regularized LS(LASSO)
LASSO_theta = partOne.LASSO_Theta(aftermat_sampleX, sampleY, 5)
LASSO_result = partOne.LASSO_predi(aftermat_polyx, LASSO_theta)
LASSO_error = partOne.error(LASSO_result, polyy)
# print('LASSO_error:', LASSO_error)
plt.figure(3)
plt.title('L1-regularized LS(LASSO)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, LASSO_result, '.-')
plt.plot(sampleX, sampleY, '*')
# Robust regression(RR)
RR_theta = partOne.RR_Theta(sampleX, sampleY, aftermat_sampleX, korder)
# print(RR_theta.shape)
RR_result = partOne.RR_predi(aftermat_polyx, RR_theta)
RR_error = partOne.error(RR_result, polyy)
# print('RR_error:', RR_error)
plt.figure(4)
plt.title('Robust regression(RR)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, RR_result, '.-')
plt.plot(sampleX, sampleY, '*')
# plt.show()
# Bayesian regression (BR)
BR_cov, BR_mean = partOne.posterior_BR(aftermat_sampleX, sampleY)
average, fangcha = partOne.BR_predi(aftermat_polyx, BR_cov, BR_mean)
BR_polyy = np.random.normal(average, np.sqrt(np.abs(fangcha)), size=None)
# print('average.shape:', average.shape)
# print('fangcha.shape:', fangcha.shape)
# print('BR_pred.shape:', BR_polyy.shape)
BR_error = partOne.error(BR_polyy, polyy)
# print('BR_error:', BR_error)
plt.figure(5)
plt.title('Bayesian regression (BR)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.legend()
plt.plot(polyx, BR_polyy, '.')
plt.plot(sampleX, sampleY, '*')
# plt.show()
return LS_error,RLS_error,LASSO_error,RR_error,BR_error
LASSO_error = partOne.error(LASSO_result, testY)
LASSO_mae_error = prt2a.mae_error(LASSO_result, testY)
print('LASSO_MSE_error:', LASSO_error)
print('LASSO_MAE_error:', LASSO_mae_error)
plt.figure(3)
plt.title('L1-regularized LS(LASSO)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.plot(testX[0], LASSO_result, '.-')
plt.plot(testX[0], testY, '.-')
# plt.plot(trainX[0], trainY,'*')
plt.legend(["test", "true"])
# Robust regression(RR)
RR_theta = partOne.RR_Theta(trainX[1], trainY, trainX, korder)
# print(RR_theta.shape)
RR_result = partOne.RR_predi(testX, RR_theta)
RR_error = partOne.error(RR_result, testY)
RR_mae_error = prt2a.mae_error(RR_result, testY)
print('RR_MSE_error:', RR_error)
print('RR_MAR_error:', RR_mae_error)
plt.figure(4)
plt.title('Robust regression(RR)')
plt.xlabel('x-value')
plt.ylabel('y-label')
plt.plot(testX[0], RR_result, '.-')
plt.plot(testX[0], testY, '.-')
# plt.plot(trainX[0], trainY,'*')
plt.legend(["test", "true"])
# Bayesian regression (BR)
BR_cov, BR_mean = partOne.posterior_BR(trainX, trainY)
average, fangcha = partOne.BR_predi(testX, BR_cov, BR_mean)