学习线性回归
用代码块功能插入代码,请勿粘贴截图
def optimier_lsm(model, X, y, reg_lambda = 0):
"""
model:模型
X:tensor 特征数据, shape = [N,0]
y:tensor 标签数据, shape = [N]
reg_lambda: 正则化系数 , 默认为0
"""
N, D = X.shape
x_bar_train = paddle.mean(X, axis=0).T
y_bar = paddle.mean(y)
x_sub = paddle.subtract(X, x_bar_train)
if paddle.all(x_sub == 0):
model.params['b'] = y_bar
model.params['w'] = paddle.zeros[shape == [D]]
return model
tmp = paddle.inverse(paddle.matmul(x_sub.T, x_sub)+
reg_lambda*paddle.eye(num_rows = [D]))
w = paddle.matmul(tmp ,paddle.matmul(x_sub.T ,y-y_bar))
b = y_bar - paddle.matmul(x_sub.T, w)
model.params['b'] = b
model.params['w'] = paddle.squeeze(w, axis=1)
return model
dimension = 1
model = Linear(dimension)
model = optimier_lsm(model, X_train.resahpe([-1,1]), y_train.resahpe([-1,1]))
print('w_pred:',model.params['w'].item(), 'b_pred',model.params['b'].item())
y_train_pred = model(X_train.resahpe(-1,1).squeeze())
train_error = mean_squared_erro(y_ture = y_train, y_pred = y_train_pred).item()
y_test_pred = model(X_test.resahpe([-1,1])).squeeze()
test_error = mean_squared_error(y_true = y_test, y_pred = y_test_pred).item()
print('test_error:',test_error)
运行结果和报错内容