Differential Privacy in Data Anonymization

This study explores an effective approach to anonymizing sensitive data, particularly focusing on diabetes-related information, through the application of Laplace noise addition. The primary objective is to protect individual privacy while preserving the utility and integrity of the dataset, especially non-numeric values. The research involves adding Laplace noise to the "Age" column to anonymize the data, subsequently evaluating the error and loss metrics by comparing the original and anonymized datasets. Key statistical measures, including mean, median, and standard deviation, are computed and contrasted between the two datasets to assess the impact of the anonymization process. Additionally, the study employs visual tools such as data distribution plots and boxplots to illustrate the differences and similarities between the original and anonymized data. The findings highlight the balance achieved between data privacy and utility, demonstrating the practical implications of privacy-preserving techniques in managing sensitive health information. The anonymized dataset is ultimately saved, showcasing a robust method for protecting privacy in sensitive datasets while maintaining their analytical value.