data_preprocess.txt

import pandas as pd
import numpy as np

# Assuming your original data is stored in a DataFrame called 'df'

# Sort the data by user ID and week number
df.sort_values(['User ID', 'Week'], inplace=True)

# Group the data by user ID
grouped = df.groupby('User ID')

# Define the sliding window size
window_size = 4

# Initialize lists to store the modified data
modified_user_ids = []
modified_item_ids = []
modified_numerical_features = []

# Iterate over each user
for user_id, group in grouped:
    # Extract the relevant columns
    user_ids = group['User ID'].values
    item_ids = group['Item ID'].values
    numerical_features = group[['Age', 'Numerical Info']].values

    # Create sliding windows
    num_windows = len(group) - window_size + 1
    for i in range(num_windows):
        modified_user_ids.append(user_ids[i:i+window_size])
        modified_item_ids.append(item_ids[i:i+window_size])
        modified_numerical_features.append(numerical_features[i:i+window_size])

# Convert the modified data to numpy arrays
modified_user_ids = np.array(modified_user_ids)
modified_item_ids = np.array(modified_item_ids)
modified_numerical_features = np.array(modified_numerical_features)

# Reshape the modified data if needed
# modified_user_ids = modified_user_ids.reshape(-1, window_size, 1)
# modified_item_ids = modified_item_ids.reshape(-1, window_size, 1)
# modified_numerical_features = modified_numerical_features.reshape(-1, window_size, num_numerical_features)

# Create a new DataFrame with the modified data
modified_data = pd.DataFrame({
    'User ID': modified_user_ids.flatten(),
    'Item ID': modified_item_ids.flatten(),
    'Age': modified_numerical_features[:, :, 0].flatten(),
    'Numerical Info': modified_numerical_features[:, :, 1].flatten()
})

# Print the modified data
print(modified_data.head())