Project for Data Protection and Privacy - Time Series Anonymization with (k,P)-anonymity

Requirements

• numpy v1.16.4
  
• pandas v0.25.0
  
• loguru v0.3.2
  
• saxpy v1.0.1.dev167
  
• matplotlib v3.1.3
  

Datasets

In the datasets folder there are two different example datasets. All dataset must be .csv and must have the columns name in the first row.
Our example datasets:

• Products.csv
  
• UrbanPopulation.csv
  

Usage

• Single test
  Run the main.py file:
  [*] Usage: python kp-anonymity.py k_value p_value paa_value max_level dataset.csv
  Where k_value must be greater than p_value, max_level must be greater than 2 and lower than 20.
  This program generates the kp-anonymization of the input dataset in a new file in the outputs folder.
  

• Multiple test
  Run the test.py file:
  [*] Usage: python test.py dataset.csv multitest
  This test runs the main.py many times and creates a .csv file containing all the average values of the NCPs (Normalized Certainty Penalty) of each table, saved in the final_table folder.
  

• Plot
  Run the test.py file:
  [*] Usage: python test.py dataset.csv plot
  This test can be done only if you have an output in final_table folder.
  This test plots the output data from multiple test step in a three-dimensional or bi-dimensional chart.
  The program chooses the pair (paa_value, max_level) that produces the best NCP. Then it plots only tuples with that values of the pair (paa_value, max_level).
  

  

Explain Parameters

• k_value: value of k-anonymity
  
• p_value: value of p-anonymity, the pattern
  
• paa_value: length of the string that can be used to describe the pattern
  
• max_level: number of letters in the alphabet used to describe the pattern