k-anonymity generalization - calculating loss

I have been given a task by my employer in which I have to anonymize a database table containing our customers' personal information so we can push out some statistics about what kind of customers buy certain goods based on various elements such as gender, age, location and more.

I've recently done some seminar work with k-anonymity and l-diversity so I accepted the task.

My employer is worried about information loss after this generalization and I assured him that I would calculate loss to keep it as low as possible without having any obvious privacy concerns.

Unfortunately I can't for the life of me remember how to calculate loss in this manner so I refereed to some examples and notes I have from the seminars.. I can't disclose any bit of our company's customer database but I have attached some screenshots of the examples from seminars.

I need someone to work out the examples and explain how lloss was calculated so I may apply the same logic to the program I am putting together to generalize our company's data.

[url removed, login to view] is the original table and [url removed, login to view] is after generalization

Programación de bases de datos Análisis estadístico Estadísticas

Nº del proyecto: #15867125

Sobre el proyecto