[Python] Getting value error when calling score and predict functions for iris dataset

class LinearRegression: def init(self, batch_size=32, regularization=0, max_epochs=100, patience=3): """Linear Regression using Gradient Descent.

Parameters:
-----------
batch_size: int
The number of samples per batch.
regularization: float
The regularization parameter.
max_epochs: int
The maximum number of epochs.
patience: int
The number of epochs to wait before stopping if the validation loss
does not improve.
"""
self.batch_size = batch_size
self.regularization = regularization
self.max_epochs = max_epochs
self.patience = patience
self.weights = None
self.bias = None

def fit(self, X, y, batch_size=32, regularization=0, max_epochs=100, patience=3):
"""Fit a linear model.

Parameters:
-----------
batch_size: int
The number of samples per batch.
regularization: float
The regularization parameter.
max_epochs: int
The maximum number of epochs.
patience: int
The number of epochs to wait before stopping if the validation loss
does not improve.
"""


# Split data into training and validation sets (90% train, 10% validation)
validation_split = 0.1
validation_size = int(len(X) * validation_split)

X_train, X_val = X[:-validation_size], X[-validation_size:]
y_train, y_val = y[:-validation_size], y[-validation_size:]
self.batch_size = batch_size
self.regularization = regularization
self.max_epochs = max_epochs
self.patience = patience

# TODO: Initialize the weights and bias based on the shape of X and y.
self.weights = None
self.bias = None
# Initialize model parameters based on shape of X and y
# n_features = X_train.shape[1] # Number of input features
n_outputs, n_features = X_train.shape
# n_outputs = y_train.shape[1] if len(y_train.shape) > 1 else 1 # Number of target variables

# Initialize weights and biases
self.weights = np.zeros(n_features) # Small random values
# self.weights = np.random.randn(n_features, n_outputs) * 0.01
# self.biases = np.zeros((1, n_outputs)) # Bias initialized to 0
self.biases = 0


# Initialize variables for early stopping
# best_weights = self.weights.copy()
# best_biases = self.biases.copy()
best_weights = []
best_biases = []
best_val_loss = float('inf')
patience_counter = 0
overall_loss = []

# TODO: Implement the training loop.
# Training Loop
for epoch in range(max_epochs):
# Shuffle the data
indices = np.random.permutation(len(X_train))
X_train_shuffled = X_train[indices]
y_train_shuffled = y_train[indices]


# Mini-batch gradient descent
for i in range(0, len(X_train), batch_size):
X_batch = X_train_shuffled[i:i + batch_size]
y_batch = y_train_shuffled[i:i + batch_size]

# Forward pass
predictions = np.dot(X_batch, self.weights) + self.biases

# Compute the error
error = predictions - y_batch

# Compute gradients for weights and biases
grad_weights = (2 / len(X_batch)) * np.dot(X_batch.T, error) + 2 * regularization * self.weights
grad_biases = (2 / len(X_batch)) * np.sum(error, axis=0, keepdims=True)

# Update parameters using gradient descent
self.weights -= 0.01 * grad_weights # learning rate = 0.01
self.biases -= 0.01 * grad_biases # learning rate = 0.01

# Evaluate the loss on the validation set after each epoch
val_predictions = np.dot(X_val, self.weights) + self.biases
val_loss = np.mean((val_predictions - y_val) ** 2) # Mean Squared Error
overall_loss.append(val_loss)

print(f"Epoch {epoch + 1}/{max_epochs}, Validation Loss: {val_loss}")

# Check for early stopping
if val_loss < best_val_loss:
best_val_loss = val_loss
best_weights = self.weights.copy()
best_biases = self.biases.copy()
patience_counter = 0 # Reset patience counter
else:
patience_counter += 1
if patience_counter >= patience:
print(f"Early stopping at epoch {epoch + 1}")
break

# Set the model parameters to the best found during training
self.weights = best_weights
self.biases = best_biases
return overall_loss



def predict(self, X):
"""Predict using the linear model.

Parameters
----------
X: numpy.ndarray
The input data.
"""
# TODO: Implement the prediction function.
X = np.array(X,dtype=float)
# print(X)
print(self.weights)

return np.dot(X, self.weights) + self.biases

def score(self, X, y):
"""Evaluate the linear model using the mean squared error.

Parameters
----------
X: numpy.ndarray
The input data.
y: numpy.ndarray
The target data.
"""
# TODO: Implement the scoring function.
# Get predictions from the model
predicted_values = self.predict(X)

# Calculate the Mean Squared Error
mse = np.mean((y - predicted_values) ** 2)
return mse


Traceback (most recent call last): File "C:\Users\kaish\PycharmProjects\PythonProject\eval_regression1.py", line 43, in mse = model.score(X_train,y_train) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\kaish\PycharmProjects\PythonProject\trial2.py", line 184, in score predicted_values = self.predict(X) ^^^^^^^^^^^^^^^ File "C:\Users\kaish\PycharmProjects\PythonProject\trial2.py", line 170, in predict return np.dot(X, self.weights) + self.biases ^^^^^^^^^^^^^^^^^^^^^^^ ValueError: data type must provide an itemsize

Tried coverting to numpy or float for self.predict but when i print that function it shows that it is string with weights and bias as values. Can someone please help me with this

Continue reading...