cookie_dp

import pandas as pd
import matplotlib.pyplot as plt

def truncate_cookie_id(cookie_id, truncation_length):
    return cookie_id[:-truncation_length]

def utility_metric(truncated_cookie_ids):
    return len(set(truncated_cookie_ids))

# Load the dataset
data = pd.read_csv('synthetic_dataset.csv')
cookie_ids = data['Cookie_ID']

# Define truncation lengths
truncation_lengths = list(range(1, len(cookie_ids[0])))

# Calculate the utility metric for each truncation length
utility_metrics = []

for truncation_length in truncation_lengths:
    truncated_cookie_ids = [truncate_cookie_id(cookie_id, truncation_length) for cookie_id in cookie_ids]
    unique_count = utility_metric(truncated_cookie_ids)
    utility_metrics.append(unique_count)

# Plot the utility vs privacy graph
plt.plot(truncation_lengths, utility_metrics)
plt.xlabel('Privacy (Truncation Length)')
plt.ylabel('Utility (Number of Unique Truncated Cookie_IDs)')
plt.title('Utility vs Privacy for Cookie_IDs in synthetic_dataset.csv')
plt.grid(True)
plt.show()