定义梯度下降函数时出现小块误差

Question

# Reading the dataset
data= pd.read_csv("energydata_complete.csv")
X = data.drop(['Appliances','date'], axis=1)
Y = data['Appliances'].values
y_np = Y.shape[0]
y = Y.reshape(y_np,1)
L = 0.001

n = X.shape[1]
m = y.size
ones = np.ones((X.shape[0],1))
X = np.concatenate((ones,X), axis = 1)
X = X.astype("float64")
y = y.astype("float64")
thetas = np.random.random([1,n+1])



# Splitting data into training and testing
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size =
0.25, random_state = 5)

# Partition the dataset in train + validation sets
from sklearn.preprocessing import MinMaxScaler
scaler = MinMaxScaler()
X_train = scaler.fit_transform(X_train)


def batch_grad_desc(X,y,m,thetas,L):
    cost_list = []
    thetas_list = []
    L = 0.001
    L = []
    m = y.size

    y_hat_list = []
    grad = True
    i = 0

    cost_list.append (1e10)

    while grad:
        y_hat = np.dot(X,thetas.T)
        y_hat_list.append(y_hat)
        Error = (y_hat - y)
        cost = 1/(2*m) * np.dot(Error.T, Error)

        cost_list.append(cost)
        thetas = thetas - (L*(1/m) * np.dot(X.T, Error))

        thetas_list.append(thetas)

        if cost_list - cost_list[i+1] < 1e-9:
            grad = False

        i+= 1
    cost_list.pop(0)

    return y_hat_list, cost_list, thetas_list


y_hat_list, cost_list, thetas_list = batch_grad_desc(X, y, m, thetas, L)
thetas = thetas_list[-1]

Error:

    TypeErrorTraceback (most recent call last)
<ipython-input-66-5b85a5574e32> in <module>
----> 1 y_hat_list, cost_list, thetas_list = batch_grad_desc(X, y, m, `thetas, L)`
      2 thetas = thetas_list[-1]

<ipython-input-65-9097ee62fbd8> in batch_grad_desc(X, y, m, thetas, L)
         19
         20         cost_list.append(cost)
    ---> 21         thetas = thetas - (L*(1/m) * np.dot(X.T, Error))
         22
         23         thetas_list.append(thetas)

TypeError: can't multiply sequence by non-int of type 'float'

I have implemented a basic batch gradient function but now I am getting error where I am updating theta and I am unable to guess how to make it work. The variables are already declared in float. I do not know what the error is all about?

Answer 1

Remove the line

L = []

You're not storing lold L 's so it's not necessary (and should be another name anyway), and I think you want L to be a float .